article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the tensors in equations and indicate a stress that is primarily biaxial in the @xmath12 plane , with principal axes closely aligned to the @xmath58 $ ] and @xmath59 $ ] crystal axes .
that there is very little tensile or compressive stress along the [ 001 ] axis is not surprising as this is the growth axis of the crystal and the nv defects measured here are @xmath60 m below the surface .
the apparent absence of any nvs oriented parallel to @xmath50 $ ] may be a result of both the stressed lattice and the nv defects having been generated during the cvd grown process , whereupon there are likely to be preferred and inhibited directions for defect orientation @xcite .
we now discuss issues related to the sensitivity of the technique . for simplicity
we have presented only the theory applicable to a stress perturbation larger than 10 mpa , equivalent to the @xmath6110 ghz splitting of the unperturbed @xmath19 level , and have treated the spin levels as degenerate .
smaller stresses can equally be determined by measuring in more detail the spectrum of the @xmath19 state and using the full theoretical description that includes the spin - orbit and spin - spin interactions @xcite .
the sensitivity is also clearly limited by the spectral resolution of the measurements , which is ultimately determined by the transition line widths .
these are a strong function of temperature , so that the sensitivity increases as the sample is cooled .
line widths approaching 10 mhz have been observed using photoluminescence excitation spectroscopy at 2.8 k @xcite , corresponding to stress magnitudes of order kpa , or strains of order 10@xmath1 .
although at such extreme sensitivity local variations in the built - in lattice stress would certainly dominate , _
changes _ in strain of this magnitude may ultimately be detectable . at higher temperatures
the line widths broaden rapidly due to the dynamic jahn teller effect @xcite , and at 77 k they are limited to about 10 ghz corresponding to stresses of order 10 mpa . at 300 k the zero phonon line is generally not clearly visible , and so stress mapping under ambient conditions would require utilisation of the stress perturbation to the electron spin state of the centre , measured using optically detected magnetic resonance ( odmr ) . the spin - orbit interaction introduces a stress component to the spin splittings in both the orbital doublet excited state @xcite and , through inter level mixing , the orbital singlet ground state @xcite . although the stress dependence of the spin transitions is much stronger in the excited state , the odmr line width is limited by the optical relaxation time ( @xmath6212 ns ) thereby limiting the sensitivity of excited state techniques .
ground state odmr probing the spin transition at 2.88 ghz is therefore likely to be the more sensitive method . in this way ,
dolde et al have demonstrated the use of single nv centres as electric field sensors with a dc sensitivity of 891 @xmath63 v @xmath64 hz@xmath65 , equivalent to a stress of about 90 kpa hz@xmath65 . by combining data from three differently oriented nv centes the full stress tensor
can similarly be extracted ( see supplementary info for details ) .
volumetric imaging of the stress tensor is possible by recording a 3-d distribution of defects . with a standard scanning confocal optical microscope
this can provide sub - micrometre spatial resolution per defect , therefore up to @xmath66 @xmath67 m resolution for the full tensor .
however nv centres in diamond are also ideally suited to super - resolution optical techniques such as stimulated emission depletion microscopy @xcite that utilise an intense depopulating beam and require highly robust photon emitters .
such techniques are entirely compatible with the methodology described here and could yield spatial resolution on a length scale of a few nanometres for systematic investigations of local stress fields .
we note that such spatial resolution would be fundamentally incompatible with the various existing approaches to stress / strain measurement .
ebsd , raman , and band gap luminescence all interrogate quantities that depend on the periodicity of the crystal lattice and can therefore only measure strain averaged over a volume , or length , that is large compared with the lattice spacing .
the use of point defects as sensors therefore opens up new possibilities for stress / strain measurement such as measuring lattice scale fluctuations in stress caused by nearby defects or surfaces , or for use in nanoscale sensors , for example in nano - electromechanical systems ( nems ) .
the sample studied was a high purity , [ 001 ] oriented type iia diamond grown by microwave plasma - assisted chemical vapour deposition ( cvd ) @xcite .
homoepitaxial cvd diamond growth was performed using a selected high pressure high temperature grown substrate , which had been carefully processed prior to growth in order to minimize the dislocation density in the grown cvd diamond . the resultant diamond had a nitrogen concentration of @xmath68 5 ppb as measured by ensemble electron paramagnetic resonance ( epr ) and a boron concentration of @xmath68 0.5 ppb as measured by secondary ion mass spectrometry ( sims ) .
after growth the sample was mechanically polished to have a surface rms roughness @xmath68 1 nm over a 1 @xmath69 1 @xmath67 m area .
low temperature photoluminescence microscopy and spectroscopy was recorded using a custom - made fiber - coupled scanning confocal microscope @xcite .
photon counting for image generation utilised a perkin elmer spcm .
spectra were recorded using an acton sp 500 monochromator and princeton spec 10:100b ccd camera .
polarization dependence of emission was measured by placing a linear optical polarizer in front of the collection aperture of the microscope .
a slight wedge in the plate leads to partial misalignment of the microscope as the polarizer is rotated , resulting in asymmetry in some of the data in figure [ polarplots ] .
the expressions for @xmath24 , @xmath25 and @xmath26 depend on the orientation of the nv centre .
since the nv axis can take one of the four @xmath70 directions ( the order of the nitrogen and vacancy sites is not important ) we require four sets of equations . to calculate these we start with the equations for the @xmath16 $ ] oriented defect with the stress tensor components referred to the crystal axes @xmath71 , @xmath72 , and @xmath52 ( see equation ) and apply successive rotations of @xmath73 around the @xmath15 axis to the coefficients matrix @xmath74 .
since the stress component @xmath40 transforms under symmetry operations as the product @xmath75 , the result of these rotations can be seen by the successive mappings @xmath76 and @xmath77 .
the full set of equations is presented below .
the use of a @xmath82 $ ] oriented sample in this study derives from the standard growth conditions of high purity cvd diamond and the ready availability of a suitable sample for study .
however as mentioned earlier this viewing axis results in an unavoidable ambiguity in the orientation of any given nv centre , effectively a mirror in @xmath52 , so that for a given nv centre we can not distinguish between @xmath16 $ ] and @xmath49 $ ] , or between @xmath50 $ ] and @xmath51 $ ] orientations . as discussed in the text , we can in some cases make good estimates of which nvs share the same orientation based on their common behaviour in a uniform stress field , but it is not possible to identify unambiguously what that orientation is .
there are two ways of avoiding this problem and thereby providing unambiguous identification of the nv orientations and thus the full stress tensor , subject only to the uniformity of the stress field .
the first is to choose a viewing axis that provides straightforward identification of the nv orientation .
for example , a @xmath16 $ ] viewing axis results in the projections of the orbital planes onto the focal plane as shown in figure [ subfigure1 ] , with four different ` ellipses ' for the four different nv orientations .
the result is that the dependence of pl intensity on excitation polarisation allows simple determination of the orientation of each nv centre .
conversely , other viewing axes are less conducive to the stress tensor evaluation .
as shown in figure [ subfigure2 ] , a @xmath83 $ ] orientation results in a twofold ambiguity in orientation as does @xmath82 $ ] .
but , more seriously , it does not allow for measurement of the angle of rotation of the @xmath84 eigenstates away from the @xmath31 axes for two of the orientations since the orbital plane is viewed edge - on .
a @xmath83 $ ] viewing axis can not therefore be used to determine the full stress tensor by this method .
the second approach is to resolve the ambiguity with a @xmath82 $ ] viewing axis , by applying a field which breaks the mirror symmetry in @xmath52 ( or in both @xmath83 $ ] and @xmath85 $ ] ) .
this could be achieved by the temporary application of an electric field , under which the different orientations would be expected to respond differently .
for instance an electric field aligned along a @xmath2 axis would allow identification of nv centres aligned parallel with that axis , as they would display a global shift of the optical transition energies but no change in splitting or polarization .
the @xmath37 electronic ground state of the nv centre has a non - zero permanent dipole as a result of a small amount of mixing with the @xmath27 symmetry excited states , which gives rise to a stress and electric field dependence of the spin transitions .
the components of this dipole aligned parallel and perpendicular to the nv axis have been measured as @xmath86 and @xmath87 @xcite .
the hamiltonian for the ground state in the presence of local stress , plus electric and magnetic fields @xcite is given by where as defined in figure [ figure1 ] , @xmath15 is along the nv axis and @xmath10 is in one of the three mirror planes .
@xmath89 is the spin - spin interaction strength , @xmath90 is the bohr magneton , @xmath91 is the electron g - factor .
@xmath92 , and @xmath93 are the components of a vector that sums the electric field and the effect of stress on a single nv centre , and @xmath94 , @xmath95 and @xmath96 are the triplet spin operators , shown below in the @xmath5 basis which result in spin eigenstates @xmath98 while @xmath99 .
matrix elements can thus be calculated using equations , , and to write the hamiltonian in the @xmath5 basis @xmath100{ccc } hd_{gs}+d_{gs}^{\parallel}\pi_z+d_{gs}^{\perp}\pi_x & -d_{gs}^{\perp}\pi_y & 0\\ -d_{gs}^{\perp}\pi_y & hd_{gs}+d_{gs}^{\parallel}\pi_z - d_{gs}^{\perp}\pi_x & 0\\ 0 & 0 & -2(hd_{gs}+d_{gs}^{\parallel}\pi_z ) \end{array } \right)\ ] ] where we have taken the case of zero magnetic field for simplicity ( we note that , as pointed out by dolde et al in @xcite , care must be taken to minimise the magnetic field , especially its @xmath15 component , to achieve high sensitivity in these measurements ) . the effect of the combined stress and electric field on the transitions between the @xmath15 spin state and the other two eigenstates is therefore described by the secular matrix in the spin
@xmath20 basis @xmath101 where @xmath102 equation above can be viewed in direct analogy with the secular matrix for the excited state as described in equation . in the absence of an electric field the same symmetry arguments that lead to equations and , expressing @xmath103 and @xmath26 for the electric dipole transitions in terms of the stress tensor components and coefficients @xmath104 , also apply here . @xmath105 and @xmath106 are therefore related in the same way to the six independent tensor components via four equivalent constants @xmath107 , @xmath108 , @xmath109 , and @xmath110 . this allows us to separate the stress contribution from the electric field contribution and write for the @xmath16 $ ] oriented centre , with stress and electric field now referred to the crystal axes , we have a set of three equations analogous to those of equation ( 2 ) , @xmath112 with the permutations for the other nv orientations as in equation . for a single nv centre , @xmath105 and @xmath106
can be measured from optically detected magnetic resonance data by the shift and splitting in the spin transitions plus the orientation of the transition - inducing microwave field .
the same pseudoinverse method can thus be applied to determine the full stress tensor from room temperature spin resonance data , once the constants @xmath107 , @xmath108 , @xmath109 , and @xmath110 have been established .
the authors would like to thank a. j. wilkinson for helpful discussions regarding stress and strain measurement techniques .
the work was funded by the united kingdom engineering and physical sciences research council through the quantum information processing interdisciplinary research collaboration ( grant reference gr / s822176/01 ) .
fg and jms acknowledge additional support from hewlett packard ltd .
30 j. f. nye , physical properties of crystals , oxford university press ( 1957 ) .
g. h. loechelt , n. g. cave , and j. menndez , polarized off - axis raman spectroscopy : a technique for measuring stress tensors in semiconductors , j. appl .
phys * 86 * , 6164 ( 1999 ) .
y. tang , d. h. rich , e. h. lingunis and n. m. haegel , polarized - cathodoluminescence study of stress for gaas grown selectively on patterned si(l00 ) , j. appl .
phys * 76 * , 3032 ( 1994 ) .
a. s. voloshin and c. p. burger , half - fringe photoelasticity - a new approach to whole field stress analysis , experimental mechanics * 23 * , 304 ( 1983 ) .
schwartz et al , electron backscatter diffraction in materials science , kluwer academic / plenum publishers , dordrecht , new york , boston , london , ( 2000 ) .
e. rittweger k. y. han , s. e. irvine , c. eggeling and s. w. hell , sted microscopy reveals crystal colour centres with nanometric resolution , nature photonics * 3 * , 144 ( 2009 ) .
a.a . kaplyanskii .
non cubic centres in cubic crystals and their piezospectroscopic investigation , optics and spectroscopy * 16 * 329 ( 1964 ) .
g. davies and m. f. hamer , optical studies of the 1.945 vibronic band in diamond , proc .
london a * 348 * , 285 ( 1976 ) . c. kurtsiefer , s. mayer , p. zarda , and h. weinfurter , stable solid - state source of single photons , phys .
85 * , 290 ( 2000 ) .
f. jelezko , t. gaebel , i. popa , a. gruber , and j.wrachtrup , observation of coherent oscillations in a single electron spin , phys .
lett . * 92 * , 076401 ( 2004 ) .
n. b. manson , j.p .
harrison , and m.j .
nitrogen - vacancy center in diamond : model of the electronic structure and associated dynamics , physical review b * 74 * 104303 ( 2006 ) .
a. batalov , v. jacques , f. kaiser , p. siyushev , p. neumann , l. j. rogers , r. l. mcmurtrie , n. b. manson , f. jelezko , and j. wrachtrup , low temperature studies of the excited - state structure of negatively charged nitrogen - vacancy color centers in diamond , phys .
102 * , 195506 ( 2009 ) .
m. w. doherty , n. b. manson , p. delaney , and l. c. l. hollenberg , the negatively charged nitrogen - vacancy centre in diamond : the electronic solution , new j. phys .
* 13 * , 025019 ( 2011 ) .
g. balasubramanian , i. y. chan , r. kolesov , m. al - hmoud , j. tisler , c. shin , c. kim , a. wojcik , p. r. hemmer , a. krueger , t. hanke , a. leitenstorfer , r. bratschitsch , f. jelezko and j. wrachtrup , nanoscale imaging magnetometry with diamond spins under ambient conditions , nature * 455 * , 648 ( 2008 ) . f. dolde , h. fedder , m.w .
doherty , t. nbauer , f. rempp , g. balasubramanian , t. wolf , f. reinhard , l. c. l. hollenberg , f. jelezko and j. wrachtrup , electric field sensing using single diamond spins , nature physics * 7 * , 459 ( 2011 ) .
a. e. hughes and w. a. runciman .
uniaxial stress splitting of doubly degenerate states of tetragonal and trigonal centres in cubic crystals .
proceedings of the physical society * 90 * 827 ( 1967 ) .
r. s. balmer , j. r. brandon , s. l. clewes , h. k. dhillon , j. m. dodson , i. friel , p. n. inglis , t. d. madgwick , m. l. markham , t. p. mollart , n. perkins , g. a. scarsbrook , d. j. twitchen , a. j. whitehead , j. j. wilman and s. m. woollard , chemical vapour deposition synthetic diamond : materials , technology and applications , j. phys . : condens .
matter * 21 * 364221 ( 2009 ) .
f. grazioso , b. r. patton , and j. m. smith , a high stability beam - scanning confocal optical microscope for low temperature operation .
review of scientific instruments * 81 * 093705 ( 2010 ) .
r.penrose and j.a . todd . on best approximate solutions of linear matrix equations .
mathematical proceedings of the cambridge philosophical society , * 52 * 17 ( 1956 ) .
p. tamarat , n. b. manson , j. p. harrison , r. l. mcmurtrie , a. nizovtsev , c. santori , r. g. beausoleil , p. neumann , t. gaebel , f. jelezko , p. hemmer and j. wrachtrup , spin - flip and spin - conserving optical transitions of the nitrogen - vacancy centre in diamond , new j. phys . * 10 * , 045004 ( 2008 ) .
u. f. s. dhaenens - johansson , a. m. edmonds , m. e. newton , j. p. goss , ; p. r. briddon , j. m. baker , p. m. martineau , r. u. a. khan , d. j. twitchen , and s. d. williams , epr of a defect in cvd diamond involving both silicon and hydrogen that shows preferential alignment , phys .
b * 82 * , 155205 ( 2010 ) .
p. tamarat , t. gaebel , j. r. rabeau , m. khan , a. d. greentree , h. wilson , l. c. l. hollenberg , s. prawer , p. hemmer , f. jelezko , and j. wrachtrup , stark shift control of single optical centers in diamond , phys .
lett * 97 * , 083002 ( 2006 ) .
c. fu , c. santori , p. e. barclay , l. j. rogers , n. b. manson , and r. g. beausoleil , observation of the dynamic jahn - teller effect in the excited states of nitrogen - vacancy centers in diamond , phys .
lett . * 103 * 256404 ( 2009 ) . g. d. fuchs , v. v. dobrovitski , r. hanson , a. batra , c. d. weis , t. schenkel , and d. d. awschalom , excited - state spectroscopy using single spin manipulation in diamond , phys .
* 101 * , 117601 ( 2008 ) .
e. van oort , and m. glasbeek , electric - field - induced modulation of spin echoes of n - v centers in diamond , chem .
. lett . * 168 * , 529 - 532 ( 1990 ) .
a. kiel and w. b. mims , linear electric field effect in paramagnetic resonance for cds : mn@xmath113 , phys .
b. * 5 * , 803 ( 1972 ) .
van oort , e. and glasbeek , m. , electric - field - induced modulation of spin echoes of n - v centers in diamond , chem .
lett . * 168 * , 529 - 532 ( 1990 ) .
f. dolde , h. fedder , m.w .
doherty , t. nbauer , f. rempp , g. balasubramanian , t.wolf , f. reinhard , l. c. l. hollenberg , f. jelezko and j.wrachtrup , electric field sensing using single diamond spins , nature physics * 7 * , 459 ( 2011 ) . | stress and strain are important factors in determining the mechanical , electronic , and optical properties of materials , relating to each other by the material s elasticity or stiffness .
both are represented by second rank field tensors with , in general , six independent components @xcite .
measurements of these quantities are usually achieved by measuring a property that depends on the translational symmetry and periodicity of the crystal lattice , such as optical phonon energies using raman spectroscopy @xcite , the electronic band gap using cathodoluminescence @xcite , photoelasticity via the optical birefringence @xcite , or electron back scattering diffraction ( ebsd ) @xcite .
a reciprocal relationship therefore exists between the maximum sensitivity of the measurements and the spatial resolution .
furthermore , of these techniques , only ebsd and off - axis raman spectroscopy allow measurement of all six components of the stress tensor , but neither is able to provide full 3d maps . here
we demonstrate a method for measuring the full stress tensor in diamond , using the spectral and optical polarization properties of the photoluminescence from individual nitrogen vacancy ( nv ) colour centres .
we demonstrate a sensitivity of order 10 mpa , limited by local fluctuations in the stress in the sample , and corresponding to a strain of about 10@xmath0 , comparable with the best sensitivity provided by other techniques . by using the colour centres
as built - in local sensors , the technique overcomes the reciprocal relationship between spatial resolution and sensitivity and offers the potential for measuring strains as small as 10@xmath1 at spatial resolution of order 10 nm .
furthermore it provides a straightforward route to volumetric stress mapping . aside from its value in understanding strain distributions in diamond , this new approach to stress and strain measurement could be adapted for use in micro or nanoscale sensors .
department of materials , university of oxford , parks road , oxford ox1 3ph , united kingdom .
school of mathematics and physics , queen s university belfast , belfast bt7 1nn , northern ireland .
tyndall national institute , university college cork , lee maltings , prospect row , cork , ireland .
element six ltd .
, king s ride park , ascot , berkshire sl5 8bp , united kingdom .
the nv centre is the most common colour centre in diamond , comprising a substitutional nitrogen atom and a vacancy in an adjacent site along one of the four @xmath2 crystal axes , as depicted in figure [ figure1](a ) .
the perturbation of its optical emission by applied uniaxial stress via the piezospectroscopic effect @xcite was characterised in the seminal work of davies and hamer in 1976 @xcite and used to identify the centre as possessing @xmath3 symmetry and a @xmath4 optical transition . in recent years , optical microscopy of individual centres has become popular since they were demonstrated to be exceptionally stable single photon emitters @xcite and to possess an electronic structure that couples the highly coherent spin state of the centre to its optical transitions , @xcite thus offering considerable potential for quantum information applications . as a result of this renewed interest ,
the electronic structure of the colour centre has been studied extensively and is now well understood in terms of the orbital energies , spin - orbit and spin - spin interactions @xcite , and hyperfine coupling .
the perturbation of the energy levels by external electric and magnetic fields is also established , leading to demonstrations of the use of nv centres as nanoscale sensors @xcite .
the nv centre structure : ( a ) schematic of the nv structure and definition of nv axes @xmath5 and crystal axes @xmath6 .
the cube represents @xmath7 of the cubic unit cell of the diamond crystal .
the nitrogen atom is colored green and the vacancy is represented by the dashed circle in the center of the cube .
the three carbon atoms adjoining the vacancy are in grey .
( b ) rotation of the @xmath8 and @xmath9 orbitals relative to nv axes @xmath10 and @xmath11 under a symmetry - breaking stress .
@xmath8 is taken to be the higher energy orbital , consistent with @xcite .
( c ) projection onto the @xmath12 image plane @xmath13 of the four possible orientations of nvs relative to the crystal axes .
( d ) projection of the @xmath14 orbitals in
( b ) onto the image plane , leading to the measured data in figure [ polarplots].,width=340 ] the effect of local stress on a single nv centre can be understood by considering the effect that different ` components ' of the stress field have on the defect symmetry . to this end
we define a set of orthogonal axes @xmath5 with @xmath15 parallel to the nv axis and @xmath10 lying in one of the three mirror planes .
for example an nv centre with its ( @xmath15 ) axis along @xmath16 $ ] we choose @xmath10 to lie along @xmath17 $ ] and @xmath11 along @xmath18 $ ] as shown in figure [ figure1](a ) .
the @xmath19 optically excited state of the nv centre has twofold spatial degeneracy and threefold spin degeneracy , and under the @xmath3 symmetry of the defect these six eigenstates are split into four levels by the spin - orbit and spin - spin interactions , with splittings of order a few ghz . under stress
the energies of the @xmath19 states are perturbed .
a stress that preserves the @xmath3 symmetry will result in a global shift of the @xmath19 levels , whilst a stress that breaks the symmetry modifies the splittings in the @xmath19 manifold and rotates the orbitals in the @xmath20 plane . in this respect the stress perturbation is similar to that of a vector with three components ( @xmath21 ) @xcite . in the limit that the stress perturbation greatly exceeds the spin - orbit and spin - spin splittings the six @xmath19 states form two branches corresponding to orthogonal spatial orbitals @xmath8 and @xmath9 , and the perturbation can be expressed as a secular matrix in the @xmath22 basis @xcite , @xmath23 where @xmath24 , @xmath25 and @xmath26 are energy shifts determined by the six components of the stress tensor .
the eigenvalues of this secular matrix are the energy changes of the stress - shifted @xmath27 states , and the eigenvectors provide the orientations of @xmath28 and @xmath29 , the @xmath30 transition dipoles , in the ( @xmath31 ) plane .
the optical transition line therefore becomes a doublet with a mean energy shift of @xmath24 and a splitting of @xmath32 , and the @xmath27 dipoles are rotated by an angle @xmath33 arctan@xmath34 relative to @xmath10 and @xmath11 ( figure [ figure1](b ) ) .
for a @xmath16 $ ] oriented nv centre the energy shifts are given by @xmath35 where the parameters @xmath36 , @xmath37 , @xmath38 , and @xmath39 have been determined as 1.47 , -3.85 , -1.04 , and -1.69 mev / gpa respectively @xcite , and the stress components @xmath40 ( where @xmath41 ) are referred to the crystal axes .
the equations for an nv centre with one of the three other orientations are obtained by performing simple rotations on the coordinate axes @xmath6 ( see supplementary information ) . the equivalent equations to with the stress referred to the nv axes are @xmath42
from which it is clear that the global shift @xmath24 results from any stress that shares the nv symmetry ( @xmath43 ) , @xmath25 results from stress that shares only the @xmath44 mirror plane symmetry ( @xmath45 ) , and @xmath26 results from stress with neither the nvs cylindrical nor its @xmath44 mirror symmetry ( @xmath46 ) .
since the stress tensor contains six independent parameters , measurement of a single nv centre can clearly provide only partial information of the local stress field . to obtain all six components of the tensor
it is therefore necessary to measure three nv centres that experience the same stress field but which have different orientations in the lattice . with @xmath24 , @xmath25 and @xmath26
determined for three differently oriented colour centres one obtains an overdetermined set of nine linear equations with six unknown variables , from which a unique solution is obtained using the moore - penrose pseudoinverse method @xcite . to demonstrate this method we present photoluminescence data from isolated nv centres in a single crystal diamond grown by plasma assisted chemical vapour deposition .
a sample was selected which displays isolated nv centres in a region of particularly high grown - in stress ( figure [ plimage ] ) , so that at a temperature of 77 k , many of the zero phonon lines reveal clear splittings of up to 0.8 nm or 2.4 mev ( figure [ plspectrum ] ) .
the viewing axis is @xmath47 $ ] , such that the integrated emission of a given nv centre is approximately @xmath48 polarized , allowing easy discrimination between nvs with axis parallel to @xmath16 $ ] or @xmath49 $ ] and those with axis parallel to @xmath50 $ ] or @xmath51 $ ] ( figure [ figure1](c ) ) .
the ambiguity ( effectively a mirror in @xmath52 ) in identifying the orientation of a given nv centre is not possible to resolve with this sample geometry , but as we shall see presents only a minor limitation to the results .
the two @xmath4 electronic transitions emit linearly polarized photons aligned with the dipole axes .
the measured projections of these polarizations onto the @xmath12 image plane allow determination of the angle @xmath53 and thus we obtain @xmath54 ) and @xmath55 .
the parameter @xmath24 is taken as the energy shift between the mid point of the two lines and 1.945 ev ( 637.5 nm ) . as the z - projections of individual nvs are not known we performed measurements on several proximal centres and looked for groupings of parameters to identify different defect orientations .
an example of a group of seven nvs studied is shown in figure [ plimage ] .
polar plots revealing the optical polarization from each of these centres are shown in figure [ polarplots ] , and the corresponding @xmath56 values listed in table [ table1 ] .
.[table1 ] polarisation and energy shift data for seven nv centres labeled in figure [ pldata ] . [ cols="<,^,^,^,^",options="header " , ] from these results we can clearly identify the @xmath16 $ ] values as relating to nv d , @xmath51 $ ] relating to nv s b and c , and @xmath50 $ ] relating to nv a. the discrepancies between the recalculated values and measured values are of order 10@xmath57 , consistent with the observed variation in parameters described previously .
noteworthy is the absence of any nvs with parameters comparable to the predictions for the @xmath49 $ ] orientation .
we attribute the orientations of all three ( more distant ) nv s e , f and g as @xmath16 $ ] the differences between their parameters and those of nv d being due to longer range changes in the stress field . |
the contact process is a famous interacting particle system modelling the spread of an infection on the sites of @xmath0 .
the evolution depends on a fixed parameter @xmath2 and is as follows : at each moment , an infected site becomes healthy at rate @xmath3 while a healthy site becomes infected at a rate equal to @xmath4 times the number of its infected neighbors .
there exists a critical value @xmath5 such that the infection , starting from the origin , infinitely expands with positive probability if and only if @xmath6 .
durrett and griffeath @xcite proved that when the contact process on @xmath0 starting from the origin survives , the set of sites occupied before time @xmath7 satisfies an asymptotic shape theorem , as in first - passage percolation . in @xcite
, two of us extended this result to the case of the contact process in a random environment .
the shape theorem can be stated as follows : provided that @xmath6 , there exists a norm @xmath8 on @xmath9 such that the set @xmath10 of points already infected before time @xmath7 satisfies : @xmath11 where @xmath12^d$ ] , @xmath13 is the unit ball for @xmath8 and @xmath14 is the law of the contact process with parameter @xmath4 , starting from the origin and conditioned to survive .
the growth of the contact process is thus asymptotically linear in time , and governed by the shape @xmath13 .
the aim of this note is to prove the continuity of the map @xmath15 .
more precisely , we prove the following result : for every @xmath6 , @xmath16 + where @xmath17 .
it is then easy to deduce the following continuity for the asymptotic shape .
denote by @xmath18 the hausdorff distance between non - empty compact sets in @xmath9 .
for every @xmath6 , @xmath19 continuity properties for asymptotic shapes in random growth models have already been investigated . in first passage percolation , perhaps the most famous random growth model , cox and kesten
@xcite proved that the time constant is continuous with respect to the distribution of the passage - time of an edge .
in a forthcoming paper , garet , marchand , procaccia and thret @xcite extend their result to the case of possibly infinite passage times by renormalization techniques .
in these two cases , thanks to a good subadditivity property , the quantity whose continuity is studied appears as an infimum of a decreasing sequence of continuous functions , which gives quite easily one half of the continuity . because of the possibility of extinction of the contact process , the subadditivity properties are not so obvious and we thus use the essential hitting time presented in garet
marchand @xcite .
note that the one - dimensional case is simpler because the growth of the supercritical contact process in dimension 1 is characterized by the right - edge velocity : its continuity is proved in liggett @xcite , theorem 3.36 .
see also durrett @xcite for an analogous result about 2d oriented percolation . in section 2
, we introduce the notation , build contact processes with distinct infection parameters on the same space thanks to the harris construction and recall the definition and properties of the essential hitting time introduced in @xcite .
section 3 is devoted to the proof of the left - continuity , while in section 4 we prove the right - continuity .
we work on the grid @xmath0 , with @xmath20 , and we put an edge between any pair of sites at distance @xmath3 for @xmath21 .
we denote by @xmath22 the set of these edges .
to define the contact process , we use the harris construction @xcite .
it allows to couple contact processes starting from distinct initial configurations and distinct parameters @xmath23 $ ] , where @xmath24 is fixed and finite , by building them from a single collection of poisson measures on @xmath25 .
we endow @xmath25 with the borel @xmath26-algebra @xmath27 , and we denote by @xmath28 the set of locally finite counting measures @xmath29 .
we endow this set with the @xmath26-algebra @xmath30 generated by the maps @xmath31 , where @xmath32 describes the set of borel sets in @xmath25 .
as the continuity is a local property , it will be sufficient in the sequel to build a coupling for contact processes with parameters in @xmath33 $ ] , for a fixed and well chosen @xmath24 .
we define the measurable space @xmath34 by setting @xmath35^{{\mathbb n}})^{{\mathbb{e}^d } } \text { and } \mathcal f=\mathcal{m}^{\otimes { \mathbb{e}^d } } \otimes \mathcal{m}^{\otimes { \mathbb{z}^d}}\otimes ( [ 0,\lambda_{\max}]^{\otimes { \mathbb n}})^{\otimes{\mathbb{e}^d}}.\ ] ] on this space , we consider the probability measure defined by @xmath36)^{\otimes { \mathbb n}})^{\otimes{\mathbb{e}^d}},\ ] ] where , for every @xmath37 , @xmath38 is the law of a poisson point process on @xmath25 with intensity @xmath4 and @xmath39)$ ] is the uniform law on the compact set @xmath40 $ ] .
fix an edge @xmath41 . denoting by @xmath42 the atoms of @xmath43
, we build the classical coupling between the poisson measures of the infection processes with different parameters @xmath44 $ ] .
define @xmath45 under @xmath46 , the random variable @xmath47 is a poisson point process with parameter @xmath4 .
we then define , for @xmath48 , the application @xmath49 the law of @xmath50 under @xmath46 is then @xmath51 we thus recover infection processes , indexed by @xmath22 , with parameter @xmath4 and recovering processes , indexed by @xmath0 , with parameter 1 .
note that the poisson measures for recoverings , @xmath52 , do not depend on @xmath4 .
the following lemma will be useful to compare the evolution of two contact processes with different parameters .
[ continuite - des - boites ] let @xmath53 and let @xmath54 be a finite subset of @xmath22 .
assume @xmath55 and note @xmath56}= \restriction{m^e_{\lambda'}}{[0,t]}\right\}.\ ] ] for each @xmath57 , there exists @xmath58 such that @xmath59 \quad | \lambda'-\lambda| \le \delta \ ; \rightarrow \ ; { \mathbb{p}}(\operatorname{idem}(s , t,\lambda,\lambda'))\ge 1-{\varepsilon}.\ ] ] let @xmath60 $ ] , and assume without loss of generality that @xmath61 . for each @xmath62 and @xmath53 , set @xmath63))=\frac{\lambda -\lambda'}{\lambda_{\max}}\lambda_{\max}t= ( \lambda'-\lambda)t.\end{aligned}\ ] ] now , @xmath64 } \ne \restriction{m^e_{\lambda'}}{[0,t ] } ) \\ & \le \sum_{e \in s}{\mathbb{p}}(d^e_t\ge 1 ) \le \sum_{e \in s}{\mathbb{e}}(d^e_t ) \le |s|t ( \lambda'-\lambda),\end{aligned}\ ] ] so we can take @xmath65 .
this construction is exposed in all details in harris @xcite ; we just give here an informal description .
suppose that @xmath66 $ ] is fixed .
let @xmath67 .
above each site @xmath68 , we draw a time line @xmath25 , and we put a cross at the times given by @xmath69 , corresponding to potential recoverings at site @xmath70 . above each edge @xmath71
, we draw at the times given by @xmath72 an horizontal segment between the extremities of the edge , corresponding to a potential infection through edge @xmath41 ( remember we fix the infection rate @xmath4 ) . an open path follows the time lines above sites but crossing a cross is forbidden and uses horizontal segments to jump from a time line to a neighboring time line : in this description , the evolution of the contact process looks like a percolation process , oriented in time but not in space . for @xmath73 and @xmath74 , we say that @xmath75 if and only if there exists an open path from @xmath76 to @xmath77 , then we define : @xmath78 for instance , we obtain @xmath79 harris proved that under @xmath46 , or under @xmath80 , the process @xmath81 is the contact process with infection rate @xmath4 , starting from initial configuration @xmath82 . for @xmath74 , we define the translation operator @xmath83 on a locally finite counting measure @xmath84 on @xmath25 by setting @xmath85 the translation @xmath83 induces an operator on @xmath86 , still denoted by @xmath83 : + for every @xmath67 , we set @xmath87)}_e)_{e\in{\mathbb{e}^d},i\ge 1 } ) .\ ] ] since the poisson point processes are translation invariant and @xmath88)$ ] is independent from the @xmath89 s , @xmath46 and @xmath80 are invariant under @xmath83 .
there is also a natural action of @xmath0 on @xmath86 , which preserves @xmath46 and @xmath80 , and which consists in changing the observer s point of view : for @xmath90 , we define the translation operator @xmath91 by setting : @xmath92 where @xmath93 the edge @xmath41 translated by vector @xmath94 . for a set @xmath95
, we define the life time @xmath96 of the process starting from @xmath82 by @xmath97 if @xmath98 , we write @xmath99 instead of @xmath100 and we simply write @xmath101 for @xmath102 .
we also note @xmath103 the critical parameter for the contact process in @xmath0 is then @xmath104 define , for @xmath6 , the following conditional probability @xmath105 for @xmath95 and @xmath90 , we also define the first infection time @xmath106 of site @xmath94 from set @xmath82 by @xmath107 if @xmath98 , we write @xmath108 instead of @xmath109 and we simply write @xmath110 for @xmath111 .
the set of points infected before time @xmath7 is then @xmath112^d.\end{aligned}\ ] ] the following estimates are classical for the supercritical contact process ; they are stated here with some extra uniformity in the parameter @xmath4 ( this uniformity is mainly obtained by stochastic comparison ) : [ propuniforme][propmoments ] let @xmath113 with @xmath114 .
there exist @xmath115 such that for every @xmath116 $ ] , for every @xmath90 , for every @xmath117 , @xmath118^d ) & \le & a\exp(-bt ) , \label{richard } \\ { \mathbb{p } } ( t<\tau_{\lambda}<+\infty ) & \le & a\exp(-bt),\label{grosamasfinis } \\ { \mathbb{p}}\left ( t_{\lambda}(x)\ge \frac{\|x\|}c+t,\ ; \tau_{\lambda}=+\infty \right ) & \le & a\exp(-bt ) .
\label{retouche } \ ] ] we now recall the definition of the essential hitting time @xmath119 . it was introduced in @xcite to prove an asymptotic shape result for the supercritical contact process in random environment .
see also garet
marchand @xcite and garet gour
marchand @xcite for further uses .
the essential hitting time @xmath119 is a time when the site @xmath94 is infected from the origin @xmath120 and also has an infinite life time .
it is defined through a family of stopping times as follows : we set @xmath121 and we define recursively two increasing sequences of stopping times @xmath122 and @xmath123 with @xmath124 : * assume that @xmath125 is defined .
we set @xmath126 . *
assume that @xmath127 is defined , with @xmath128 .
we set @xmath129 .
we then set @xmath130this quantity represents the number of steps before we stop : either we stop because we have just found an infinite @xmath131 , which corresponds to a time @xmath132 when @xmath94 is occupied and has infinite progeny , or we stop because we have just found an infinite @xmath133 , which says that after @xmath131 , site @xmath94 is never infected anymore .
when the contact process survives , the second case does almost surely not occur .
in @xcite , using and , it is proved that @xmath134 is almost surely finite , which allows to define the essential hitting time @xmath119 by setting @xmath135 it is of course larger than the hitting time @xmath110 and can been seen as a regeneration time . at the same time
, we define the operator @xmath136 on @xmath86 by : @xmath137 the advantage of the essential hitting time @xmath119 is that it enjoys , unlike @xmath110 , some good invariance and integrability properties when conditioned to survive .
we now recall the main results of @xcite we will need here . in the following , we fix @xmath138 such that @xmath114 .
@xmath139 * for each @xmath6 , for every @xmath90 , @xmath140 * there exist constants @xmath141 such that for every @xmath116 $ ] , for every @xmath90 , for every @xmath142 ,
@xmath143 \le c_p ( 1+\|x\|)^{p}.\ ] ] * for each @xmath6 , for every @xmath90 , there exists a deterministic @xmath144 such that @xmath145 the convergence holds @xmath14 almost surely , and also in @xmath146 . *
the function @xmath147 can be extended to a norm on @xmath9 .
let @xmath148 * for every @xmath57 , @xmath149 , for every @xmath7 large enough , @xmath150 it has also been noted in @xcite ( see proof of lemma 25 there ) that there exists @xmath151 such that , for each @xmath116 $ ] and each @xmath152 , the sequence @xmath153 is subadditive .
thus , with , we can represent @xmath144 as the following infimum : @xmath154 \quad \forall x \in { \mathbb{z}^d}\quad \mu_{\lambda}(x)={\renewcommand{\arraystretch}{0.6 } \begin{array}{c } { \scriptstyle } \\
\inf\\ { \scriptstyle n\ge 1 } \end{array } \renewcommand{\arraystretch}{1}}\frac{m_1+{\overline{\mathbb{e}}}_{\lambda}(\sigma_{\lambda}(nx))}n.\ ] ] as a corollary of , we obtain the following monotonicity property : [ coromonotonie ] for each @xmath155 , @xmath156 is non - increasing on @xmath157 .
suppose @xmath158 .
choose @xmath113 with @xmath159 .
use the construction of subsection [ subs : constr ] to build the two contact processes with respective parameters @xmath4 and @xmath160 . on the event @xmath161 , which has positive probability
, we have that for each @xmath162 , @xmath163 . letting @xmath164 go to infinity
, we get @xmath165 .
we prove here the left - continuity of @xmath166 .
more precisely , we prove that for each @xmath167 , for every @xmath168 , there exists @xmath169 such that @xmath170 \quad \forall x\in \mathbb s^{d-1}\quad |\mu_{\lambda_0}(x)-\mu_{\lambda}(x)|\le \varepsilon.\ ] ] when proving continuity theorems for the time constant in first passage percolation ( see cox and kesten @xcite ) , the left - continuity is usually considered as the easy part , due to the fact that the time constant is an infimum . in the case of the contact process , there are extra difficulties , because contact processes with different intensities can not be coupled in such a way that they die simultaneously .
[ unemonotonie ] let @xmath6 . for each @xmath90 ,
@xmath171 fix @xmath6 .
choose @xmath172 such that @xmath173 and set @xmath174 .
fix @xmath90 .
use the construction of subsection [ subs : constr ] . in this proof , for @xmath175 $ ] , we note @xmath176 .
suppose that for every @xmath175 $ ] , we have managed to construct a `` good '' event @xmath177 such that @xmath178 on @xmath177 .
then , for @xmath175 $ ] , @xmath179 now , using the fact that @xmath180 and the control on the moments of @xmath181 , we get @xmath182 thus if we prove that @xmath183 and @xmath184 are close to @xmath3 when @xmath160 is close to @xmath4 , we can complete the proof .
we now build the `` good '' event @xmath177 such that @xmath178 on @xmath177 and such that @xmath184 goes to @xmath3 as @xmath160 goes to @xmath4 . since @xmath185 is @xmath14-a.s .
finite and @xmath186 is @xmath14-a.s . a finite set , we can first choose @xmath187 such that @xmath188^d\ } \cap \{\sigma_{\lambda}\le m\}. \ ] ] then , estimates and let us choose @xmath189 such that for each @xmath190 $ ] @xmath191^d\}\cap \{l<\tau_{\lambda'}<\infty\}^c.\ ] ] set @xmath192^d\cap { \mathbb{z}^d}$ ] and @xmath193 .
with lemma [ continuite - des - boites ] , we can choose @xmath169 such that @xmath194 \quad { \overline{\mathbb{p}}}_{\lambda}(\operatorname{idem}(s , t,\lambda,\lambda'))\ge 1-{\varepsilon}/3.\ ] ] finally , we consider , for every @xmath195 $ ] , the event @xmath196 the choices , and we respectively made for @xmath197 and @xmath198 , and the invariance property ensure that @xmath199 the event @xmath200 says that , apart from the fact that @xmath201 , the time @xmath185 is determinated by the configuration of the poisson processes in the space - time box @xmath202^d\times [ 0,m]$ ] .
the event @xmath203 says that , if @xmath204 has a progeny for parameter @xmath160 still alive at time @xmath205 , then @xmath201 .
the event @xmath206 says that the infection at rate @xmath160 in the box @xmath207 $ ] behaves exactly like the infection at rate @xmath4 in the same box .
now , on the event @xmath208 , the point @xmath204 has a progeny for parameter @xmath4 that is still alive at time @xmath205 .
but the infections with rate @xmath4 and @xmath160 coincide in the box @xmath207 $ ] , so the point @xmath204 has a progeny for parameter @xmath160 that is also still alive at times @xmath205 .
then @xmath209 , and it is now easy to see that @xmath178 .
note also that @xmath210 .
this gives @xmath211 and , coming back to , we see that @xmath212 \quad { \overline{\mathbb{e}}}_{\lambda'}(\sigma_{\lambda ' } ) & \le \frac1{1-{\varepsilon } } \left ( { \overline{\mathbb{e}}}_{\lambda}(\sigma_{\lambda})+\sqrt{c_2(1+\|x\|^2 ) } \sqrt{{\varepsilon}}\right).\end{aligned}\ ] ] this completes the proof
. [ lemmgauche ] for each @xmath155 , @xmath156 is left - continuous on @xmath157 .
fix @xmath90 .
since , from corollary [ coromonotonie ] , the application @xmath156 is non - increasing on @xmath157 , we can define @xmath213 obviously @xmath214 and we must prove @xmath215 . note @xmath216 . using the representation of @xmath217 as an infimum
, we have @xmath218 by lemma [ unemonotonie ] , for each @xmath219 , @xmath220 , so @xmath221 which completes the proof .
the difference between ( [ e : contgauche ] ) and lemma [ lemmgauche ] is the uniformity of the control . for all @xmath222 ,
since @xmath8 is a norm and by symmetry of the model , we have for all @xmath223 , @xmath224 where @xmath225 .
we obtain that @xmath226 .
fix @xmath227 and @xmath228 . by lemma [ lemmgauche ]
we know that @xmath229 , thus there exists @xmath230 such that for all @xmath231 $ ] , for all @xmath232 , we have @xmath233 . we obtain the existence of @xmath234 such that for all @xmath232 satisfying @xmath235 , we have @xmath236 } \ { | \mu_\lambda ( x ) - \mu_\lambda ( y ) | \ } \leq \varepsilon \,.\ ] ] there exists a finite set of points @xmath237 in @xmath9 such that @xmath238 thus for all @xmath231 $ ] we obtain @xmath239 by homogeneity of @xmath8 , the result of lemma [ lemmgauche ] also holds for all @xmath240 , in particular for @xmath241 .
this concludes the proof of ( [ e : contgauche ] ) . we can notice that the previous argument also applies to the study of the right - continuity of @xmath166
however , as we will see in the next section , we do not need it since we perform directly the study of the right - continuity of @xmath166 uniformly in all directions .
we prove here the right - continuity of @xmath166 .
more precisely , we prove that for each @xmath167 , for every @xmath168 , there exists @xmath169 such that @xmath242 \quad \forall x\in\mathbb{s}^{d-1}\quad |\mu_{\lambda_0}(x)-\mu_{\lambda}(x)|\le \varepsilon.\ ] ] as we will see , the right - continuity of the asymptotic shape of the contact process can be obtained by a slight modification of a part of the proof of the large deviations inequality for the contact process established by garet and marchand in @xcite .
let @xmath244 and @xmath245 be positive integers .
consider @xmath246 and close to @xmath247 .
we define the following event , relative to the space - time box @xmath248^d\cap { \mathbb{z}^d})\times[0,2n]$ ] : @xmath249-ln , ln[^d\right\}.\end{aligned}\ ] ] consider first @xmath250 .
the first part of the event ensures that the descendants , at time @xmath251 , of any point @xmath252 in the box @xmath253 are included in @xmath254 : it is a sharp control , requiring the asymptotic shape theorem for parameter @xmath247 .
the second part ensures that the descendants , at all times in @xmath255 $ ] , of the whole box @xmath253 are included in @xmath256-ln , ln[^d$ ] : the bound is rough , only based on the ( at most ) linear growth of the process with parameter @xmath247 .
thus , the `` good growth '' event @xmath250 is typical , and it has been proved that garet and marchand used lemma [ catendvers12 ] to prove the upper large deviations for the contact process : for every @xmath262 $ ] , provided that @xmath263 is fixed as in lemma [ catendvers12 ] , then for @xmath264 greater than some @xmath265 , they prove that there exists @xmath266 such that @xmath267 the idea of the proof is as follows : a too fast infection from @xmath268 to @xmath269 uses a too fast path , along which we should find a number of order @xmath270 of `` bad growth '' events , _
i.e. _ translated versions of @xmath271 .
the proof ends with a peierls argument : the event @xmath271 is local , thus its translated events are only locally dependent . if their probability is small enough , we can control the probability that there exists a path from @xmath268 to @xmath269 with at least @xmath270 `` bad growth '' events .
let s come back to the right - continuity .
fix @xmath167 and @xmath57 .
take @xmath272 given by lemma [ catendvers12 ] , @xmath264 large enough , and @xmath273 as before
. a look at the proof of the peierl argument in @xcite should convince the reader that for each @xmath274 , we have @xmath275 remember that the event @xmath276 is local .
thus , applying lemma [ continuite - des - boites ] with @xmath277^d\cap { \mathbb{z}^d}$ ] and @xmath278 , we obtain the existence of @xmath279 $ ] such that @xmath280\quad{\mathbb{p}}(a_{\lambda,\lambda_0}^{\alpha , l , n,{\varepsilon}/3})>p_1.\ ] ] then , it follows that @xmath280\quad \exists a , b\quad \forall t>0\quad { \mathbb{p}}(\xi^{0,\lambda}_t \not \subset ( 1+{\varepsilon})t{\mathcal{s}(\lambda_0)})\le a\exp(-b t).\ ] ] now , we can deduce from and that @xmath280\quad \exists a , b\quad \forall t>0\quad { \overline{\mathbb{p}}}_{\lambda}(h^{0,\lambda}_t \not \subset ( 1+{\varepsilon})t{\mathcal{s}(\lambda_0)})\le a\exp(-b t).\ ] ] a detailed proof is provided in @xcite when deducing ( 62 ) from ( 61 ) . fix @xmath281 $ ] , @xmath282 and , using the asymptotic shape result , choose @xmath7 large enough to have @xmath283 and @xmath284 .
this implies that the event @xmath285 has positive probability ; particularly , @xmath286 , and , letting @xmath287 tend to @xmath120 , we have @xmath288 \quad { \mathcal{s}(\lambda)}\subset ( 1+{\varepsilon}){\mathcal{s}(\lambda_0)},\ ] ] or equivalently @xmath288 \quad\forall x\in{\mathbb{r}^d}\quad \mu_{\lambda_0}(x)\le ( 1+{\varepsilon})\mu_{\lambda}(x).\ ] ] this completes the proof of . | we prove the continuity of the shape governing the asymptotic growth of the supercritical contact process in @xmath0 , with respect to the infection parameter .
the proof is valid in any dimension @xmath1 . |
a _ word _ is a finite sequence of letter chosen from a finite alphabet @xmath3 .
the periodicity of words is a classical and well - studied topic in both discrete mathematics and combinatorics on words , starting with the classic paper of fine and wilf and continuing with the works of guibas and olydzko . for more recent work ,
see , for example , .
we say that a word @xmath4 has period @xmath5 if @xmath6 = w[i+p]$ ] for all @xmath7 that make the equation meaningful .
( if @xmath8 and one indexes beginning at position @xmath9 , this would be for @xmath10 . )
trivially every word of length @xmath11 has all periods of length @xmath12 , so we restrict our attention to periods @xmath13 .
the least period is sometimes called _ the _ period . for example , the french word entente has periods @xmath14 , @xmath15 , and @xmath16 . empirically , one quickly discovers that a randomly chosen word typically has a least period that is very close to its length .
this readily follows from the fact that the number of words over a given alphabet grows exponentially as the length increases .
it can also be seen as a particular case of the fact that most strings are not compressible . in this paper
, we quantify this basic observation and show that the expected least period of a string of length @xmath11 over an @xmath1-letter alphabet is @xmath17 , where @xmath18 is @xmath19 .
another concept frequently studied in formal language theory is that of _ border _ of a word @xcite .
a word @xmath20 has border @xmath4 if @xmath4 is both a prefix and a suffix of @xmath20 .
normally we do not consider the trivial borders of length @xmath21 or @xmath22 .
thus , for example , the english word ionization has one border : ion .
less trivially , the word alfalfa has two borders : a and alfa .
a word with no borders is _
unbordered_. there is an obvious connection between periods of a word and its borders : if @xmath4 has a period @xmath5 , then it has a border of length @xmath23 .
for example , the english word abracadabra , of length @xmath24 , has periods @xmath16 , @xmath25 , and @xmath24 , while it has borders of length @xmath9 and @xmath26 .
consequently , the least period of a word corresponds to the length of the longest border ( and an unbordered word corresponds to least period @xmath11 , the length of the word ) .
the reader should be constantly aware of this duality , since it is often useful and more natural to think about periods in terms of borders .
this can be seen from the announced result : it is more compact to speak directly about the expected maximum border length , which is @xmath18 .
if @xmath27 is a set of integers , we shall write @xmath28 for @xmath29 , and @xmath30 for @xmath31 . by @xmath32 , we mean the prefix of length @xmath7 of the word @xmath33 .
we shall obtain our results by counting words with a given length @xmath11 and a given finite set of periods @xmath34 , or equivalently , with a given set of border lengths @xmath28 . for technical reasons , in order to be able to deal with unbordered words , we shall always suppose that @xmath35 , that is , we shall say that every word has a border length zero .
there are two basic types of requirements .
let @xmath36 and let @xmath37 be the cardinality of @xmath38 .
similarly , let @xmath39 and let @xmath40 be the cardinality of @xmath41 .
words with many periods have been amply studied . in particular
, there is a fast algorithm constructing a word of length @xmath11 with periods @xmath27 and maximal possible number of letters .
such a word , called an @xmath42-word in the literature , is unique up to renaming of the letters .
let @xmath43 denote the cardinality of the alphabet of the @xmath42-word of length @xmath11 and periods @xmath27 .
let @xmath44 and @xmath45 . the well - known periodicity lemma ( often called the fine and wilf theorem , which is the origin of the term @xmath42-word ) states that if a word of length at least @xmath46 has periods @xmath5 and @xmath47 , then it also has period @xmath48 .
moreover , the bound @xmath46 is sharp ; for all @xmath49 there are words of length @xmath50 with period @xmath5 and @xmath47 but not period @xmath48 .
this can be stated , using the just - introduced terminology , by the two assertions @xmath51 and @xmath52 .
the number @xmath43 can be computed and the corresponding @xmath42-word constructed using the algorithm of tijdeman and zamboni @xcite ( see @xcite for an alternative presentation ) .
the computation is summarized by the following formula : @xmath53 where @xmath54 and @xmath55 . since each word having the periods in @xmath27 ( and possibly others ) results from a coding ( a letter - to - letter mapping ) of the corresponding @xmath42-word
, we obtain @xmath56 which is the starting point of our computation .
note that @xmath57 is the set of words from @xmath58 having least period @xmath5 .
equivalently , @xmath59 is the set of words with the longest border of length @xmath60 . for @xmath61 ,
let @xmath62 denote the relative number of such words .
our goal is to compute @xmath63 which is the expected maximum border length for words in @xmath58 .
we first show that this quantity converges as @xmath11 approaches infinity .
[ limit ] for each @xmath64 and each @xmath61 , @xmath65 case 1 : @xmath66 . then @xmath67 recall that @xmath68 ( resp . ,
@xmath69 ) counts the words with longest border length @xmath60 from @xmath70 ( resp . , @xmath58 ) .
first , note that @xmath71 for any @xmath5 and @xmath11 .
this implies @xmath72 and we are done .
case 2 : @xmath73 .
there is a useful correspondence between @xmath58 and @xmath70 , given by the insertion of a letter in the middle of the shorter word .
the basic observation , already used in @xcite , is that this insertion does not influence borders of length at most @xmath74 .
define @xmath75 then @xmath76 .
let @xmath77 and write @xmath78 with @xmath79 , @xmath80 , and @xmath81 .
the words @xmath4 and @xmath82 have the same borders up to length @xmath74 .
since @xmath83 , we deduce that @xmath82 has a border of length at least @xmath84 , that is , a period at most @xmath85 .
this implies @xmath86 similarly , a word @xmath87 has period at most @xmath88 , and so @xmath89 we thus obtain @xmath90 for each @xmath64 and @xmath91 , the limits @xmath92 exist .
furthermore , the convergence is exponential
. follows directly from the definition of @xmath93 and lemma [ limit ] .
from the estimates of the previous section , we know that @xmath93 and @xmath94 both converge quickly to @xmath95 and @xmath96 , respectively .
thus , they can be estimated to a few digits by explicit enumeration . in order to evaluate @xmath18 to dozens of decimal places ,
however , we need a more efficient way to calculate @xmath97 .
this can be done using the recurrence formulas that we derive below .
they are reformulations and generalizations of formulas given by harborth @xcite for sets of periods .
we first prove the following auxiliary claim . [ pq ]
let a word @xmath4 have a period @xmath98 and let @xmath99 be the prefix of @xmath4 of length @xmath23 .
then @xmath4 has a period @xmath100 if and only if @xmath99 has a period @xmath101 .
note that @xmath99 is a border of @xmath4 .
the following conditions are easily seen to be equivalent : * @xmath4 has a period @xmath47 , * @xmath4 has a border of length @xmath102 , * @xmath99 has a border of length @xmath102 , * @xmath99 has a period @xmath103 . since @xmath104
, the proof is completed .
[ rec ] let @xmath27 be a set of periods with @xmath54 and @xmath105 .
then @xmath106 where @xmath107 \ell^{2p - n}\cdot { { f}}_\ell(p - p , n - p ) , & \text{if $ p\geq { \left\lceil n/2 \right\rceil}$ } \ , . \end{cases}\end{aligned}\ ] ] from @xmath37 we have to subtract the number of words from @xmath58 that have periods @xmath27 but also a period smaller than @xmath108 .
we define , for each @xmath109 , the set @xmath110 if @xmath111 then @xmath112 is empty , since a word @xmath113 also has a period @xmath114 , and @xmath115 contradicts the definition of @xmath116 .
moreover , the sets @xmath117 are pairwise disjoint , and @xmath118 it remains to show that @xmath119 is the cardinality of @xmath112 for each @xmath120 .
let @xmath121 .
we claim that @xmath122 is a one - to - one mapping of @xmath112 to @xmath123 .
let @xmath113 .
by lemma [ pq ] , the word @xmath124 has periods @xmath125 and no period @xmath126 with @xmath127 , that is , no period less than @xmath128 . since @xmath129 and
since @xmath124 also has a period @xmath5 , we have @xmath130 .
similarly , one can verify that if @xmath131 , then @xmath132 and @xmath133 .
let @xmath134 .
again , using lemma [ pq ] , it is straightforward to verify that @xmath135 if @xmath136 is small , then we can formulate a more explicit formula that uses the mbius @xmath137-function . let @xmath27 be a set of periods with @xmath138 .
then @xmath139 let @xmath4 be a word of length @xmath11 with a period @xmath108 and let @xmath5 be the least period of @xmath4 .
then , by the periodicity lemma , we have that @xmath5 divides @xmath108 , since @xmath140 implies @xmath141 .
therefore , for each divisor @xmath5 of @xmath108 , @xmath142 and the claim follows from mbius inversion .
in this section we derive explicit formulas for @xmath143 and @xmath144 , which are the asymptotic probabilities that a random word is unbordered , or has longest border of length one , respectively .
these are two cases in which lemma [ rec ] yields a relatively simple expression , since @xmath145 .
the number of unbordered words satsifies a well known recurrence formula ( see , e.g. , ( * ? ? ?
* , eq . ( 34 ) ) for the binary case and @xcite for the general case ) .
the formula can be verified using theorem [ rec ] but we shall give an elementary proof . in this section ,
let @xmath146 denote @xmath147 , and let @xmath148 denote @xmath149 .
[ unb ] @xmath150 for @xmath151 , the verification is straightforward
. let @xmath20 and @xmath152 be nonempty words with @xmath153 and consider words @xmath154 , @xmath155 and @xmath156 where @xmath157 and @xmath158 are letters .
since the shortest border of @xmath155 has length at most @xmath159 , the word @xmath154 is unbordered if and only if @xmath155 is .
this proves @xmath160 if @xmath11 is odd . on the other hand
, @xmath156 can have the shortest border of length @xmath161 .
therefore , @xmath156 is unbordered if and only if ( i ) @xmath154 is unbordered and ( ii ) @xmath162 . since the shortest border is itself unbordered , we obtain @xmath163 if @xmath11 is even .
theorem [ unb ] directly yields , for each @xmath164 , @xmath165 therefore @xmath166 defining the generating function @xmath167 , we get @xmath168 the next step is to obtain a functional equation for @xmath169 : @xmath170 therefore @xmath171 successively substituting @xmath172 , @xmath173 , @xmath174 , , we get @xmath175 since it is easy to see that @xmath176 we obtain @xmath177 a similar analysis was given previously by @xcite , although our analysis is slightly cleaner .
there is also a relatively simple recurrence for @xmath178 , that is , for words with the longest border of length 1 .
the particular case @xmath179 was previously given by harborth ( * ? ? ? * , eq .
( 36 ) ) . in this section
, we let @xmath180 denote @xmath178 , and let @xmath181 denote @xmath182 .
[ bor1 ] @xmath183 verify that @xmath184 and @xmath185 , and let @xmath20 and @xmath152 be nonempty words with @xmath153 .
consider words @xmath186 , @xmath187 and @xmath188 where @xmath189 are ( not necessarily distinct ) letters .
the letter @xmath190 is the longest border of the word @xmath187 if and only if ( i ) @xmath190 is the longest border of @xmath186 and ( ii ) @xmath191 .
moreover , ( i ) @xmath190 is the shortest border of @xmath186 , and ( ii ) @xmath192 @xmath193 if and only if @xmath190 is the shortest border of @xmath194 .
this implies @xmath195 for @xmath196 odd .
similarly , @xmath190 is the shortest border of @xmath188 if and only if ( i ) @xmath190 is the longest border of @xmath186 and ( ii ) @xmath197 . as above
, we have to subtract the number of words @xmath194 with the longest border @xmath190 .
it follows that @xmath198 for @xmath199 even . from theorem [ bor1 ]
, we deduce @xmath200 using , we obtain @xmath201 defining the generating function @xmath202 , we then get @xmath203 a functional equation for @xmath204 is obtained as follows : @xmath205 we have @xmath206 and @xmath207 from here , we deduce @xmath208 we do not know how to obtain similar expressions for other border lengths .
theorem [ rec ] , as well as explicit formulas from the previous section allow fast computer evaluation of @xmath93 and @xmath209 for large @xmath11 , and therefore also evaluation of @xmath210 and @xmath95 with high precision .
we list some rounded values in the following tables .
@xmath211 @xmath212 and some values of @xmath213 rounded to four decimal digits : @xmath214 for example , we see that a long binary word chosen randomly has about @xmath215 chance to be unbordered .
a bit more probable , at @xmath216 , is that such a word will have its longest border of length one . over a five - letter alphabet , more than three words out of four
are unbordered , on average .
figure [ fig1 ] shows the distribution of lengths of the shortest period for binary words of length @xmath217 . , width=432 ] [ fig1 ] our original motivation was a question about the average period of a binary word .
the answer is , that the border of a binary word has asymptotically constant expected length , namely @xmath218
recently there has been some interest in computing the expected value of the largest unbordered factor of a word .
this is a related , but seemingly much harder , problem .
a. loptev , g. kucherov , and t. starikovskaya . on
maximal unbordered factors . in f.
cicalese et al . ,
editor , _ cpm 2015 _ , volume 9133 of _ lecture notes in computer science _ , pages 343354 .
springer - verlag , 2015 . | we investigate the behavior of the periods and border lengths of random words over a fixed alphabet .
we show that the asymptotic probability that a random word has a given maximal border length @xmath0 is a constant , depending only on @xmath0 and the alphabet size @xmath1 .
we give a recurrence that allows us to determine these constants with any required precision .
this also allows us to evaluate the expected period of a random word . for the binary case ,
the expected period is asymptotically about @xmath2 .
we also give explicit formulas for the probability that a random word is unbordered or has maximum border length one . |
in the nonet of pseudoscalar mesons the most interesting one for the investigation of the gluon sector of qcd is the relatively heavy @xmath0 , which is related to the so - called @xmath1 problem @xcite . since the work of thooft @xcite there is little doubt that the large @xmath0 mass has its origin in the gluon sector of qcd .
the singlet @xmath2 , which is the main component of the @xmath0 ( apart from admixtures of the octet meson @xmath4 ) , is expected to couple directly to the gluons via the gluon anomaly and gives an additional mass term for the @xmath0 .
an important ingredient here is the mixing angle which has been determined rather accurately by experiments via the analyses of the decays of the @xmath5 and @xmath0 @xcite during the last years .
there are different approaches to calculate the mass of the @xmath0 . in their pioneering works
witten @xcite and veneziano and di vecchia @xcite constructed a meson lagrangian which includes the gluon anomaly .
as a byproduct witten derived a formula which relates the mass of the @xmath2 to the topological susceptibility .
witten s formula has been a key tools for theoretical investigations using rather detailed models of the qcd gluon vacuum such as the instanton model @xcite or the monopole condensate model @xcite . recently
hutter @xcite was able to relate the topological susceptibility to the gluon condensate in the simple picture of the gluon vacuum as an ensemble of uncorrelated instantons and anti - instantons and obtained a good estimate of the mass of the @xmath0 .
also recently the model of the squeezed gluon vacuum has been considered as an interesting alternative @xcite-@xcite to the above approaches . in the present paper
we discuss the squeezed vacuum in a simple variant and apply it to the @xmath1 problem .
we shall directly relate the anomaly term to the gluon condensate .
numerical results for the gluon condensate are compared with other values given by shifman , vainsthein and zakharov @xcite and also recently by narison @xcite .
the paper is organized as follows : in section 2 we briefly recall the @xmath1 problem and discuss its connection to the gluon condensate . in section 3 the effective low energy meson lagrangian which includes an anomalous gluon term is quoted and in section 4 the witten formula is discussed . in section 5
we outline the squeezed vacuum . in section 6 the mass of the @xmath0
is related to the squeezed gluon condensate and numerical results are given . finally our conclusions are drawn .
the pseudoscalar mesons have been well understood as irreducible representations of the flavour su(3 ) . the pions and kaons are members of the octet representation , whereas the @xmath5 and @xmath0 mesons are related to the octet and singlet pseudoscalar states @xmath6 via the mixing @xmath7 with a mixing angle @xmath8 .
different experimental data have been used to determine the mixing angle @xmath8 .
recent analyses of @xmath9 and @xmath0 decays @xcite have obtained a mixing angle @xmath10 the experimental values for the masses of the @xmath5 and the @xmath0 are @xcite @xmath11 > from these experimental values of @xmath12 and @xmath8 one obtains the masses @xmath13 and @xmath14 of the @xmath4 and @xmath2 @xmath15 on the other hand the masses @xmath13 and @xmath14 can be obtained from the following gell - mann okubo formulae @xcite @xmath16 which have been derived using the gell - mann oakes renner relations for the nonet of pseudoscalar goldstone bosons under the assumption of only small explicit chiral @xmath17 symmetry breaking by the @xmath18 , @xmath19 and @xmath20 quark masses . here
@xmath21 mev is the pion decay constant and @xmath22 @xcite is the singlet decay constant .
taking the kaon and pion masses from experiment , this relation gives @xmath23 , which is in reasonable agreement with ( [ m88 ] ) would lead to complete agreement .
this deviation from the experimental value ( [ phiexp ] ) could have its origin in a small deviation from gell - mann okubo formula which can lead to a relatively large change in the mixing angle . ] .
for the @xmath2 mass , however , the gell - mann okubo formula yields only a value @xmath24 for @xmath25 , which is much smaller than the experimental ( [ m00 ] ) .
the large squared difference @xmath26 shows that the small explicit chiral symmetry breaking by the current quark masses alone can not account for the large mass of the @xmath2 or the @xmath0 .
this constitutes the well - known @xmath1 problem @xcite . on the way towards the solution of this puzzle
it is important to note that even in the chiral limit of vanishing quark masses the colour singlet axial @xmath27 quark current @xmath28 is actually not conserved on the quantum level but afflicted with an anomaly due to the gluon sector of qcd @xcite @xmath29 with @xmath30 where @xmath31 is the gluon field strength tensor , @xmath32 is the diagonal matrix of the current quark masses , @xmath33 the strong coupling constant and @xmath34 the number of light quark flavours .
although derived only in the one - loop approximation in the presence of classical background gluon fields it is generally accepted that ( [ anomaly ] ) is actually an operator identity ( see e.g. the discussion in @xcite ) . the anomalous term @xmath35 is known as pontryagin density .
an important condition for a large @xmath0 mass is that there is a possibility for a nonvanishing @xmath35 as will be discussed in the following .
an important concept in this context is the gluon condensate , defined as the expectation value of the local gluonic operator @xmath36 in the nonperturbative qcd vacuum @xcite @xmath37 approximate empirical values for the physical gluon condensate are the estimate @xmath38 by shifman , vainshtein and zakharov @xcite and the update average value @xmath39 obtained by narison @xcite in a recent analysis of heavy quarkonia mass - splittings in qcd .
the value of @xmath40 in the low energy region is not known very well from experiment .
the value used by shifman , vainshtein and zakharov @xcite is @xmath41 and that used by narison @xcite in the low energy region is @xmath42 .
a nonvanishing value of the gluon condensate can lead to a nonvanishing anomalous density @xmath35 and therefore to a large mass of the @xmath0 . as discussed by hutter @xcite for a dilute noninteracting gas of statistically independent instantons and anti - instantons , which are ( anti-)selfdual field configurations in euclidean space with @xmath43 ,
the local gluonic operator @xmath44 defined in ( [ gcop ] ) is proportional to the sum of the number densities of instantons and anti - instantons .
the pontryagin density @xmath35 given in ( [ cpd ] ) , on the other hand , is equal to the difference of the number densities of instantons and anti - instantons in the dilute instanton gas . in the model of the qcd vacuum as a non - interacting ensemble of instantons and anti - instantons
the existence of a gluon condensate is therefore a necessary condition for a nonvanishing value of @xmath35 and hence for a large mass of the @xmath0 .
furthermore it has been suggested in @xcite and will be discussed in more detail in this paper that also a squeezed condensate in minkowski space can lead to a large mass of the @xmath0 through the @xmath1 anomaly .
there are several ways to implement the gluon anomaly in calculations of the meson spectrum in order to obtain the large value of the @xmath0 mass . in ref .
@xcite thooft introduced an effective quark interaction in minkowski space simulating the anomalous term which breaks @xmath1 but conserves the chiral @xmath45 symmetry .
this determinantal interaction has been widely used within effective quark models such as the njl model @xcite .
dorokhov and kochelev @xcite model the @xmath0 as a mit bag where the nonperturbative vacuum of qcd is allowed to enter the bag which leads to instanton induced quark interactions .
they obtain the values @xmath46 and @xmath47 .
the pions obtained in the same scenario , however , turn out to be much too heavy .
for the calculation of the mass of the @xmath0 mass in microscopic models of the gluon vacuum , such as the magnetic monopole condensate , the instanton gas model or the squeezed condensate , another quite general approach is very convenient and will be discussed in the next two sections .
it is based on witten s formula derived from a low energy meson lagrangian which contains the axial @xmath1 anomaly at tree level .
in a quite general framework , without using the concept of instantons , the authors @xcite-@xcite start from the low energy effective glueball - meson lagrangian in a general @xmath48 vacuum @xmath49 + \frac{1}{2}v { \rm tr } [ m_q(u^+ + u ) ] \nonumber\\ & + & \frac{i}{2}{\cal q}{\rm tr } [ \ln
u -\ln u^+ ] + \frac{n_f}{af_0 ^ 2}{\cal q}^2 -\theta { \cal q}~,\end{aligned}\ ] ] where @xmath50~,\end{aligned}\ ] ] with the octet meson fields @xmath51 @xmath52\end{aligned}\ ] ] and the singlet @xmath2 . the pseudoscalar glueball field @xmath53 plays the role of an auxilary field included into the meson lagrangian in order to implement the axial @xmath1 anomaly on the hadronic level .
it can be identified with the vacuum expectation value of the pontryagin density @xmath35 defined in ( [ cpd ] ) in terms of gluon degrees of freedom as will be discussed further below .
the vacuum angle @xmath48 , the parameters @xmath54 and @xmath55 , as well as the diagonal mass matrix of the current quarks @xmath56 are to be fixed by comparison with experiment
. measurements of the dipole moment of the neutron @xcite limit the vacuum angle to @xmath57 , so that in practice it can be taken equal to zero .
the kinetic term reads explicitly @xmath58 & = & -{1\over 2}\partial_{\mu}\pi^0\partial^{\mu}\pi^0 -\partial_{\mu}\pi^+\partial^{\mu}\pi^- -\partial_{\mu}k^+\partial^{\mu}k^- \nonumber\\ & & -\partial_{\mu}k^0\partial^{\mu}\bar{k}^0 -{1\over 2}\partial_{\mu}\eta_8\partial^{\mu}\eta_8 -{1\over 2}\partial_{\mu}\eta_0\partial^{\mu}\eta_0\end{aligned}\ ] ] in terms of the pion , kaon and eta fields .
the explicitly chiral symmetry breaking mass term is @xmath59 & = & { 1\over 2 } v\bigg [ { 1\over 2}(m_u+m_d)(\pi^0\pi^0 + 2\pi^+\pi^- ) \nonumber\\ & & + ( m_u+m_s)k^+\bar{k}^- + ( m_d+m_s)k^0\bar{k}^0 \nonumber\\ & & + { 1\over 6}(m_u+m_d+4m_s)\eta_8 ^ 2 + { f_{\pi}^2\over 3 f_0 ^ 2}(m_u+m_d+m_s)\eta_0 ^ 2\nonumber\\ & & + ( m_u - m_d){1\over \sqrt{3}}\pi^0 ( \eta_8 + { \sqrt{2}f_{\pi}\over f_0}\eta_0)\nonumber\\ & & + { \sqrt{2}f_{\pi}\over 3f_0}(m_u+m_d-2m_s)\eta_8\eta_0 \bigg]~. \label{mesonmasses}\end{aligned}\ ] ] the corresponding meson mass formulae implicit in ( [ mesonmasses ] ) are the gell - mann oakes renner relations . they can be combined in order to yield the gell - mann okubo relations ( [ gmo1 ] ) and ( [ gmo2 ] ) . since @xmath60 & = & { \sqrt{2n_f}\over f_0 } \eta_0~,\end{aligned}\ ] ] only the singlet field @xmath2 is coupled to @xmath53 .
the euler - lagrange equations for @xmath61 and @xmath62 of the meson lagrangian ( [ u ] ) include the chiral anomaly @xmath63 + 2n_f { \cal q}(x)\end{aligned}\ ] ] with the axial u(1 ) current @xmath64 $ ] from noethers theorem .
( [ anomalyb ] ) is the hadron analogue to ( [ anomaly ] ) with the pseudoscalar glueball field identified as the vacuum expectation value of the pontryagin density @xmath35 defined in ( [ cpd ] ) in terms of gluon degrees of freedom .
the lagrangian ( [ u ] ) therefore includes the chiral anomaly . using @xmath65 to eliminate the auxilary pseudoscalar glueball field @xmath53 leads to the reduced lagrangian @xmath66 + \frac{1}{2}v{\rm tr}[m_q(u^+ + u ) ]
\nonumber\\ & & -\frac{af_0 ^ 2}{4 n_f}\left(\theta-\frac{i}{2}{\rm tr } [ \ln
u -\ln u^+]\right)^2~.\end{aligned}\ ] ] noting ( [ eta_0u ] ) we see that the @xmath2 field attains an additional contribution @xmath67 to its mass from the gluon anomaly , which can be chosen in accordance with ( [ dm ] ) .
a different derivation of the reduced lagrangian ( [ ured ] ) , which contains the @xmath1 anomaly at tree level , has been given by witten @xcite using large @xmath68 arguments .
note that the kinetic term and the mass term in the lagrangian ( [ ured ] ) are of order @xmath69 in the number @xmath68 of colours , since @xmath70 , whereas the third anomalous term like the mass @xmath54 due to the gluon anomaly itself are of order @xmath71 and therefore vanish in the large @xmath68 limit .
> from the anomalous low energy meson lagrangian ( [ ured ] ) one can derive the quite general formula by witten @xcite , which allows one to calulate the mass shift of the @xmath2 due to the gluon anomaly in leading order in @xmath72 for microscopic models of the gluon vacuum .
although @xmath48 is practically zero , witten proposed to use its fluctuations to calculate the quadratic mass shift @xmath73 in the following way . from ( [ ured ] ) and ( [ dma ] ) one obtains @xmath74 here @xmath75 is the ground state energy density of the hamiltonian corresponding to ( [ ured ] ) with no quarks , i.e. all meson fields set equal to zero .
the singlet decay constant @xmath76 has been replaced by @xmath77 which is in agreement with experiment @xcite and in accordance with large @xmath68 arguments . as noted by witten ,
the ground state energy @xmath78 can be alternatively considered in pure gluon qcd with the lagrangian including an anomalous term as @xmath79 with @xmath80 @xmath81 and @xmath82 are the components of the chromoelectric and chromomagnetic field strength . for the quantization in the hamilton formalism
we use the weyl gauge @xmath83 , such that @xmath84 . introducing the canonical momenta
@xmath85 the hamiltonian reads @xmath86~.\end{aligned}\ ] ] the hamiltonian contains a term linear in @xmath48 @xmath87 and a term quadratic in @xmath48 @xmath88 in order to calculate the topological susceptibility @xmath89 it is only necessary to calculate the vacuum energy @xmath78 up to second order in @xmath48 . for this purpose
one has to do second order perturbation theory in the operator @xmath90 and only first order in @xmath91 . this way witten @xcite derived an expression for the topological susceptibility of qcd without quarks : @xmath92 where @xmath93 is the interaction picture gluon vacuum corresponding to @xmath94 .
the subscript conn denotes the connected part of the green function and @xmath95 is the dyson time ordering operator .
( [ etam ] ) and ( [ witfo ] ) together constitute the witten formula .
the first term in ( [ witfo ] ) is a propagator term whereas the second one is a contact term .
the contact term can be incorporated into the propagator term , if instead of the dyson @xmath95 ordering the wick @xmath96 ordering is used @xcite . in the following
we shall use ( [ witfo ] ) with the dyson @xmath95 ordering .
witten kept only the propagator term in ( [ witfo ] ) and dropped the contact term .
he argued that although the contact term is necessary for the positivity of the result , it should not contribute to the mass of the @xmath0 since no goldstone pole can appear in a one point function .
based on witten s formulae ( [ etam ] ) and ( [ witfo ] ) , dropping the contact term , there were several approaches to calculate the large mass of the @xmath0 .
as one of the earliest approaches novikov et al .
@xcite used the euclidean model of the gluon vacuum as an ensemble of noninteracting ( anti-)instantons . in this context
is valid @xmath97 with @xmath35 and @xmath44 given in ( [ cpd ] ) and ( [ gcop ] ) and @xmath98 denoting the dilute instanton gas vacuum .
he then related the right hand side of ( [ qqnn ] ) to the gluon condensate using the ward identity @xcite @xmath99 corresponding to the anomalous breaking of scale invariance . inserting ( [ qqnn ] ) and
( [ nnn ] ) into witten s formulae ( [ etam ] ) and ( [ witfo ] ) in euclidean space , he obtained the following relation between the mass of the @xmath0 and the physical value of the gluon condensate @xmath100 @xmath101 the coefficient @xmath102 , originating from the fermion determinant in the qcd path integral , accounts for the suppression of the gluon condensate value due to the presence of light quarks @xmath103 a discussion of the value for @xmath104 has been given by novikov et al .
@xcite in the instanton gas scenario .
they find values in the range @xmath105 .
recently hutter @xcite showed that formula ( [ metanov ] ) for the mass of the @xmath0 remains valid even for the more general case of the gluon vacuum as an ensemble of noninteracting instantons and anti - instantons . using the somewhat wider range @xmath106 , hutter obtains a value @xmath107 neglecting the quark masses .
another application of the above witten formula has been the calculation of the mass of the @xmath0 in the magnetic monopole condensate scenario by ezawa and iwazaki @xcite .
using the hypothesis of dominance of the abelian gauge field components at large distances they find a value @xmath108 neglecting the influence of quarks on the gluon condensate . a further interesting possibility to explain the large mass of the @xmath0 , which has been suggested recently @xcite is the model of a squeezed gluon vacuum to be discussed in the following two sections .
the squeezed condensate of gluons has been investigated recently @xcite-@xcite in order to construct a lorentz and gauge invariant stable qcd vacuum in minkowski space .
different alternative approaches have not solved this problem .
for instance the simple perturbative vacuum is unstable @xcite , and there is no stable ( gauge invariant ) coherent vacuum in minkowski space @xcite .
> from the physical point of view , the squeezed state differs from the coherent one by the condensation of colour singlet gluon pairs rather than of single gluons . in analogy to the bogoliubov model
@xcite we consider the case of a homogeneous condensate , but in a squeezed instead of a coherent state . let @xmath109 denote the eigenstates of the pure gluon hamiltonian ( [ htheta ] ) for vanishing vacuum angle @xmath110 . in order to introduce the squeezed states let us consider the gluon system to be enclosed in a large finite volume @xmath111 .
the squeezed states @xmath112\rangle ~$ ] as candidates for the gluon eigenstates @xmath109 , in particular the squeezed vacuum @xmath113\rangle ~$ ] as a candidate for a homogeneous colourless gluon vacuum @xmath114 , are constructed from the nonperturbative states @xmath115\rangle\big|_{\xi=0}$ ] , further specified below , according to @xmath116 & = & \exp\left[i{\xi\over 2}v({\cal a}^a_{i } { \cal e}^a_{i}+ { \cal e}^a_{i}{\cal a}^a_{i})\right]\end{aligned}\ ] ] with the zero momentum components @xmath117 and @xmath118 of the fields and their canonical momenta contains the parameter @xmath119 given below .
this special transformation for the homogeneous condensate does not violate lorentz invariance , since the gauge fields are massless @xcite .
the question of gauge invariance of such a procedure is a difficult open problem and first steps towards a clarification are under current investigation @xcite . as in ref .
@xcite we suppose here the gauge invariance of @xmath120 and hence of the squeezing operator as a colourless functional of the spatial zero momentum components of the gauge fields .
the multiplicative transformations of fields corresponding to ( [ sqt1 ] ) and ( [ sqt2 ] ) are @xmath121~{\cal a}_i^a~u_{\rm sq}^{-1}[\xi ] & = & { \rm e}^{\xi } { \cal a}_i^a~ , \nonumber\\ \vspace{0.2 cm } u_{\rm sq}[\xi]~{\cal e}_i^a~u_{\rm sq}^{-1}[\xi ] & = & { \rm e}^ { - \xi } { \cal e}_i^a~.\end{aligned}\ ] ] > from this canonical transformation it follows that the expectation values in the squeezed state basis as functions of the squeezing parameter @xmath119 behave like @xmath122|\left({\cal b}_i^a\right)^2 & = & { \rm e}^{4\xi}\langle{}n^{(0)}|\left({\cal b}_i^a\right)^2|n'^{(0 ) } \rangle ~~ , \label{b2sq}\end{aligned}\ ] ] @xmath122|\left({\cal e}_i^a\right)^2 { \rm e}^{-2\xi}\langle{}n^{(0)}|\left({\cal e}_i^a\right)^2 @xmath122|{\cal e}_i^a { \cal b}_i^a { \rm e}^{\xi}\langle{}n^{(0)}|{\cal e}_i^a { \cal b}_i^a @xmath123 .
let the reference states @xmath124 be such that the expectation values @xmath125 , @xmath126 and @xmath127 behave in the large volume limit ( @xmath128 ) like @xmath129 in accordance with dimensional analysis .
the parameter of the squeezing transformation @xmath119 can be chosen so that the magnetic condensate density ( [ b2sq ] ) remains finite in the large volume limit ( @xmath130 ) @xmath131 .
this entails that the electric component ( [ e2sq ] ) and the mixed component ( [ ebsq ] ) of the gluon condensate vanish in the large volume limit @xmath132 ) the gluon condensate is equal to its magnetic part , @xmath133 of full qcd can be related to the gluon condensate @xmath100 via the one - loop result for the scale anomaly in the dilatation current @xmath134 here @xmath135 is the vacuum expectation value of the trace of the energy momentum tensor and @xmath136 the qcd @xmath137-function coefficient . since the vacuum energy @xmath138 is negative and @xmath139 positive , the gluon condensate @xmath100 is expected to be positive and hence @xmath140 which is a lorentz invariant statement and corresponds to a magnetic vacuum .
quark condensate terms have been neglected in this estimation and are expected to soften the inequality ( [ magnele ] ) but not to turn it to the reverse .
the squeezed vacuum has vanishing electric field @xmath141 and thus is in accordance with ( [ magnele ] ) .
for the calculation of the @xmath0 mass in the squeezed vacuum it is useful to rewrite the witten formula ( [ witfo ] ) in the schrdinger picture as @xcite @xmath142 it contains the exact eigenstates @xmath109 and eigenvalues @xmath143 of the pure gluon qcd hamiltonian .
we approximate ( [ witfos ] ) by replacing the exact eigenstates and eigenvalues by the nonperturbative squeezed states and the corresponding energy expectation values . in order to see whether the propagator and the contact term give finite contributions we inspect their volume dependence . since in the squeezed vacuum @xmath144 the contact term gives a finite contribution to the topological susceptibility according to ( [ b ] ) .
in addition a further , negativ contribution might arise from the propagator term .
despite the fact that the matrix elements are suppressed in the large volume limit the denominator can simultaneously become very small due to states arbitrarily close to the vacuum .
whereas in the instanton model of the gluon vacuum only the propagator term is considered , as discussed in section 4 , we shall here not further investigate the propagator term but consider only the finite contribution from the contact term .
the contact term by itself gives the following contribution to the @xmath2 mass via ( [ etam ] ) @xmath145 ) for the squeezed gluon condensate and relation ( [ lq ] ) to account for the suppresion of the physical gluon condensate due to the presence of light quarks by a factor @xmath102 we obtain @xmath146 this formula is the main result of our investigation .
it relates the gluon condensate to the @xmath1 breaking contact contribution to the mass shift of the @xmath2 .
we point out that nielsen et al .
@xcite have derived the same formula for the shift of the @xmath0 mass as ( [ meta0 ] ) ( except for the factor @xmath104 ) using the cheshire cat principle instead of the witten formula .
a main source of uncertainty is the reduction factor of the gluon condensate due to the presence of light quarks .
we shall use here the estimate @xmath147 obtained by novikov et al .
@xcite in the instanton gas scenario .
the value of @xmath40 in the low energy region is not known very well from experiment .
the value used by shifman , vainshtein and zakharov @xcite is @xmath41 and that used by narison @xcite in the low energy region is @xmath42 . in order to check whether relation ( [ meta0 g ] )
is in agreement with empirical data we have plotted in fig .
1 the gluon condensate @xmath148 against @xmath40 for two limiting values @xmath149 and @xmath150 of @xmath104 . we estimate the contact contribution by the full empirical mass shift @xmath151 obtained in ( [ dm ] ) for @xmath152 . the curves in fig .
1 give our result for these two limiting values @xmath153 ( solid line ) and @xmath154 ( dashed line ) and the mixing angle @xmath152 .
the thin lines indicate the error @xmath155 in the value of the mixing angle which is negligibly small . also shown
are the gluon condensate value @xmath38 by shifman , vainshtein and zakharov @xcite ( filled square ) and the update average value @xmath39 for the gluon condensate obtained by narison @xcite ( filled triangle ) in a recent analysis of heavy quarkonia mass - splittings in qcd .
the gluon condensate values described by our result ( [ meta0 g ] ) are in good agreement with both the shifman , vainshtein and zakharov and the narison value for the respective values of @xmath40 in the range of narison s update average @xmath40 ( filled circle ) . compared with our previous result @xcite
we have included in this work a rather detailed discussion of the error due to the experimental uncertainty in the mixing angle and due to the influence of the quarks on the gluon condensate .
in the present work we have pointed out the possibility to resolve the @xmath1 problem via the model of a homogeneous squeezed gluon condensate as an interesting alternative to existing models of the qcd gluon vacuum such as the instanton gas model . in particular we have discussed that in the squeezed vacuum the contact term in witten s formula can give a sizable contribution to the gluonic part of the @xmath2 mass . in the framework of a homogeneous squeezed vacuum
we have obtained a relation between the value of the gluon condensate and the mass shift of the @xmath2 as a function of the strong coupling constant .
an interesting aspect to consider the contact term is the fact that h.b .
nielsen et al .
@xcite have derived exactly the same relation using the cheshire cat principle istead of witten s formula .
the gluon condensate values found in our estimate are in quite good agreement with both the `` standard '' value @xmath156 by shifman , vainsthein and zakharov and the update average value @xmath157 by narison for reasonable values of the strong coupling in the low energy region . in our simple model of the squeezed vacuum
only the zero momentum mode of the gluon field operators has been squeezed .
this is a lorentz invariant operation since the gauge field is massless .
the question of the gauge invariance of the procedure is still an open problem and under current investigation @xcite as well as the possibility of a lorentz and gauge invariant extension to the squeezing of nonzero momentum modes .
we thank d. ebert , i.i .
kogan , e.a .
kuraev and c.d .
roberts for fruitful discussions on the subject .
p.p . is grateful to the deutsche forschungsgemeinschaft for support under contract no .
ro 905/11 - 1 .
m.k.v . acknowledges financial support provided by intas under grant no .
w 94 - 2915 and by the max - planck gesellschaft as well as the hospitality of the fachbereich physik at the university of rostock .
99 s. weinberg , phys .
d 11 ( 1975 ) 3583 ; s. weinberg , the quantum theory of fields ii ( cambridge university press , new york,1996 ) , ch .
g. t hooft , phys .
( 1976 ) 8 ; phys .
d 14 ( 1976 ) 3432 ; 18 ( 1978 ) 2199 ( e ) ; phys .
142 ( 1986 ) 357 .
pham , phys .
b 246 ( 1990 ) 175 .
p. ball , j .-
frre and m. tytgat , phys .
b 365 ( 1996 ) 367 .
e. witten , nucl .
b 156 ( 1979 ) 269 .
g. veneziano , nucl .
b 159 ( 1979 ) 213 ; p. di vecchia , phys . lett .
b 85 ( 1979 ) 35 ; p. di vecchia and g. veneziano , nucl . phys .
b 171 ( 1980 ) 253 . v.a .
novikov , m.a .
shifman , a.i .
vainshtein and v.i .
zakharov , nucl .
b 191 ( 1981 ) 301 .
t. schfer and e.v .
shuryak , instantons in qcd , hep - ph/9610451 .
z.f . ezawa and a. iwazaki , phys .
d 26 ( 1982 ) 631 . m. hutter , phys .
b 367 ( 1996 ) 275 . l.s .
celenza and c.m .
shakin , phys .
d 34 ( 1986 ) 1591 . t.s .
biro , ann .
( ny ) 191 ( 1989 ) 1 ; phys .
b 278 ( 1992 ) 15 ; int .
phys . 2 ( 1992 ) 39 .
a. mishra , h. mishra , s.p .
misra and s.n .
nayak , phys .
d 44 ( 1991 ) 110 ; z. phys .
c 37 ( 1993 ) 233 ; a. mishra , h. mishra and s.p .
misra , z. phys .
c 57 ( 1993 ) 241 ; z. phys .
c 59 ( 1993 ) 159 .
i.i . kogan and a. kovner ,
d 52 ( 1995 ) 3719 .
d. blaschke , h .-
pavel , v.n .
pervushin , g. rpke and m.k .
volkov , phys .
b 397 ( 1997 ) 129 .
shifman , a.i .
vainshtein and v.i .
zakharov , nucl .
b 147 ( 1979 ) 385 , 448 , 519 .
s. narison , phys .
b 387 ( 1996 ) 162 . review of particle physics , phys .
d 54 ( 1996 ) .
shifman , phys .
209 ( 1991 ) 341 .
s. klimt , m. lutz , u. vogl and w. weise , nucl .
phys . a 516 ( 1990 ) 429 ; + s.p .
klevansky , rev .
phys . 64 ( 1992 ) 649 ; + t. hatsuda and t. kunihiro , phys .
247 ( 1994 ) 221 .
v. dmitrasinovi , phys .
c 53 ( 1996 ) 1383 and references therein .
a.e . dorokhov and n.i .
kochelev , sov .
52 ( 1 ) ( 1990 ) 135 ; sov .
23 ( 5 ) ( 1992 ) 522 .
crewther , phys .
lett . 70b ( 1977 ) 349 . c. rosenzweig , j. schechter and c.g .
trahern , phys .
d 21 ( 1980 ) 3388 .
p. nath and r. arnowitt , nucl .
b 209 ( 1982 ) 251 ; r. arnowitt and p. nath , phys . rev .
d 25 ( 1982 ) 595 .
volkov , sov .
nuclei 13 ( 1982 ) 446 , 17 ( 1986 ) 186 .
cheng , phys .
158 ( 1988 ) 1 .
dyakonov and m.i .
eides , sov .
jetp 54 ( 1981 ) 232 .
savvidy , phys .
b 71 ( 1977 ) 133 .
h. leutwyler , nucl .
b 179 ( 1981 ) 129 .
bogoliubov , j. phys . 11 ( 1947 ) 23 .
a. linde , particle physics and inflationary cosmology ( harwood academic publishers , amsterdam , 1990 ) , p. 69 .
gogilidze , a.m. khvedelidze , d.m .
madlenov and h .-
pavel , hamiltonian reduction of su(2 ) dirac - yang - mills mechanics , rostock report mpg - vt - ur 109/97 , dubna report e2 - 97 - 217 , hep - th/9707136 .
d. schtte , phys .
d 31 ( 1985 ) 810 .
shuryak , the qcd vacuum , hadrons and the superdense matter ( world scientific , singapore , 1988 ) , ch . 1.9 .
nielsen , m. rho , a. wirzba and i. zahed , phys .
b 281 ( 1992 ) 345 . | the relation between the large mass of the @xmath0 and the structure of the gluon vacuum via the @xmath1 anomaly is discussed .
a squeezed gluon vacuum is considered as an alternative to existing models . considering witten s formula for the @xmath2 mass
we show that the contact term can give a sizable contribution and relate it to the physical gluon condensate .
the values of the gluon condensate obtained through this relation are compared with the value by shifman , vainshtein and zakharov and the recent update values by narison .
pacs number(s ) : 12.38.aw , 12.40.yx , 14.40.aq , 14.70.dj squeezed vacuum , gluon condensate , u@xmath3(1 ) symmetry breaking , @xmath2 mass formula and |
a number of systems , including moving vortex lattices in type - ii superconductors with disordered pinning @xcite , sliding charge - density waves @xcite , and driven wigner crystals @xcite , can be modeled as moving lattices interacting with random quenched disorder .
if the external driving force is weak the particles are pinned , while at high external drives the particles can slide .
there is extensive evidence that in the sliding phase the particles organize into a moving anisotropic crystal or moving smectic phase and that the motion is confined to well defined channels @xcite .
a remarkable property of these channels is that if an additional external drive is applied in the _ transverse _ direction the moving system exhibits a finite transverse depinning threshold @xcite .
this transverse depinning threshold has been confirmed in a number of simulations @xcite and experiments @xcite .
a transverse depinning barrier is also present for particles moving over a periodic substrate . in this type of system ,
the transverse velocity force curves exhibit a devil s staircase structure as the particle motion locks to the symmetry directions of the periodic substrate @xcite .
this effect occurs even for a single particle moving over a periodic substrate @xcite .
in contrast , a single particle moving over a random substrate will not show a transverse depinning threshold or a devil s staircase response . in ref .
@xcite , a finite transverse depinning force was predicted for a vortex lattice , and it was also conjectured that a moving lattice would align with the direction of the initial longitudinal drive in order to minimize power dissipation . if the moving lattice remains aligned in this initial orientation as a transverse drive is applied , the transverse velocity force curves could exhibit a devil s staircase response as the applied force vector locks with the triangular symmetry of the moving lattice .
in contrast to a system with a periodic substrate , here the symmetry breaking which leads to the preferred directions of motion comes from the moving lattice itself . numerically observing the devil s staircase in a system with random disorder
can be difficult since large systems and long time averages of the transverse velocity are needed to reduce the fluctuations .
an added difficultly with observing this type of devil s staircase is that fine increments of the applied transverse force are required to resolve the higher order steps .
another possibility for the dynamics at the transverse depinning transition which has not been considered is that the moving lattice structure could become strongly disordered at the depinning transition , leading to plastic distortions .
if such a _ plastic transverse depinning _ occurs , it is natural to ask if the particles would reorder into a tilted channel structure , causing a new transverse barrier to develop .
to address these issues we consider two types of two - dimensional ( 2d ) systems with periodic boundary conditions .
the first is a vortex system containing @xmath0 vortices and @xmath1 short range random pinning sites .
the vortices are modeled as point particles and their dynamics evolve via an overdamped equation of motion , given for vortex @xmath2 as @xmath3 where @xmath4 is the damping constant .
the vortex - vortex interaction force @xcite is @xmath5 , where @xmath6 is the modified bessel function , @xmath7 is the distance between vortices @xmath2 and @xmath8 located at @xmath9 and @xmath10 , @xmath11 , the unit of force is @xmath12 , @xmath13 is the flux quantum , and the london penetration depth @xmath14 is the unit of length .
this interaction is appropriate for stiff 3d vortices in relatively thick films where the film thickness @xmath15 is much larger than @xmath14 but is smaller than the larkin - ovchinnikov correlation length @xcite .
time is measured in units of @xmath16 and a molecular dynamics ( md ) time step is equal to @xmath17 .
we find identical results if smaller time steps are used . for a 0.1@xmath18 m thick crystal of nbse@xmath19 , @xmath20n / m and @xmath21 ns / m @xcite , so that @xmath22s .
the pinning force @xmath23 arises from pinning sites which are modeled as attractive parabolic potentials of radius @xmath24 and strength @xmath25 : @xmath26 , where @xmath27 is the distance between vortex @xmath2 and pin @xmath28 located at @xmath29 , @xmath30 , and @xmath31 is the heaviside step function .
the pinning interaction is intended to represent columnar pinning @xcite .
the net external driving force @xmath32 is comprised of the longitudinal force applied in the @xmath33-direction , @xmath34 , and the transverse force applied in the @xmath35-direction , @xmath36 .
the vortex ( pin ) density is given by @xmath37 ( @xmath38 ) .
the system is prepared by simulated annealing , after which a longitudinal force is gradually applied in increments of @xmath39 up to a specific value .
the final longitudinal driving force is held fixed and then the transverse drive is applied in small increments .
we measure the transverse velocity @xmath40 , where @xmath41 is the velocity of vortex @xmath2 .
we also measure the fraction of sixfold coordinated vortices , @xmath42 , where @xmath43 is the coordination number of vortex @xmath2 as determined from a voronoi construction .
the second system we consider is sliding wigner crystals composed of @xmath44 electrons .
here we use the same langevin type simulations which were employed previously to identify the sliding states in this system @xcite . the electron - electron interaction force @xmath45 , where @xmath46 and @xmath47 is the location of election @xmath48 .
the electron crystal has lattice constant @xmath49 .
the disorder comes from @xmath1 positively charged impurities placed at a distance @xmath50 out of plane with the form @xmath51 , where @xmath52 here @xmath53 is the in - plane projection of the location of pin @xmath8 .
the number of electrons @xmath44 equals the number of impurities @xmath1 . both the electron - electron and electron impurity interactions are long range , and we evaluate them with a fast - converging sum @xcite . to measure the transverse response ,
we follow a similar procedure as in the vortex system .
we first consider a vortex system for a sample of size @xmath54 with @xmath55 , @xmath56 , and fixed longitudinal drive @xmath57 .
for this value of @xmath58 and pinning density we find three regimes of the transverse response for varied pinning strength .
for @xmath59 the system is in what we term the weak pinning limit , and the equilibrium state at @xmath60 is free of dislocations . at @xmath61 ,
there is a finite transverse depinning threshold and the transverse depinning occurs without the generation of additional defects . in the intermediate regime @xmath62 , the equilibrium @xmath63 state contains dislocations , and for @xmath64
the transverse depinning is plastic and is accompanied by the generation of defects . under a longitudinal drive , for @xmath65 the vortices dynamically reorder into a defect - free lattice , while for @xmath66 the vortices undergo a partial dynamical reordering into a moving smectic state containing a finite number of dislocations that are all aligned with the driving direction , as seen in previous simulations @xcite .
the number of defects in the moving smectic state gradually increases with increasing @xmath25 until each row of vortices is completely decoupled from the neighboring rows .
a weak smectic state contains only a few dislocations so that several adjacent rows are coupled but rows separated by longer length scales are decoupled . in the strong pinning regime @xmath67 ,
the vortices flow plastically through the system and the moving channel structure is destroyed so that the dislocations are no longer aligned with the drive . in this case
there is no transverse depinning barrier .
we have tested system sizes up to @xmath68 and we find only dislocation free lattices in the equilibrium state for @xmath69 .
it is possible that for much larger systems than we can simulate , dislocations will eventually appear in the weak pinning systems .
vs the transverse applied force @xmath70 for a vortex lattice moving over random disorder of strength @xmath71 with fixed longitudinal driving force @xmath72 .
( b ) @xmath73 vs @xmath70 for the same system .
( c ) @xmath74 vs @xmath70 showing the evolution of the @xmath75 step for @xmath76 , 0.04 , 0.06 , and @xmath77 , from top to bottom .
( d ) step widths @xmath78 in @xmath79 vs @xmath25 for the @xmath80 , @xmath81 , @xmath75 , @xmath82 , @xmath83 and @xmath84 steps , from bottom to top.,width=336 ] we consider a sample with @xmath71 in the weak pinning regime where the equilibrium lattice contains no dislocations .
we fix the longitudinal drive to @xmath61 and increase @xmath70 from zero in increments of @xmath85 every @xmath86 md time steps . in order to resolve the steps in the transverse response , we average the transverse velocity @xmath74 over @xmath87 md steps .
the appearance of the curve does not change if @xmath74 is averaged over @xmath88 md steps instead .
a typical simulation runs for a total of more than @xmath89 md steps .
when the net driving force vector aligns with a symmetry direction of the vortex lattice , we find that the transverse response locks to that direction over a range of @xmath70 . for a moving triangular lattice ,
this locking occurs whenever @xmath90 where @xmath91 and @xmath92 are integers .
in such a devil s staircase , the most pronounced steps occur when both @xmath91 and @xmath92 have low values .
the plot of @xmath74 vs @xmath70 in fig .
[ fig : image](a ) shows that a transverse critical depinning threshold @xmath93 occurs at @xmath94 .
it is important to note that the vortex lattice remains aligned in the @xmath33-direction during the entire simulation . for increasing @xmath70 ,
a number of steps appear in @xmath74 corresponding to the locking of the net applied force vector with a symmetry direction of the lattice .
a prominent step occurs near @xmath95 which corresponds to the case of @xmath96 .
the step at @xmath97 corresponds to the ( 1,3 ) state , and the ( 2,3 ) state appears at @xmath98 .
there are also a number of smaller steps at higher values of @xmath91 and @xmath92 , confirming the existence of a devil s staircase response . in fig .
[ fig : image](b ) , @xmath73 vs @xmath70 more succinctly highlights the steps .
the peaks in fig .
1(b ) correspond to points at which the curvature in @xmath74 vs @xmath70 increases just before and after a plateau region .
the most pronounced steps occur for @xmath99 where the value of both @xmath92 and @xmath91 are small , which is typical for a devil s staircase structure .
this result confirms the prediction of giamarchi and le doussal @xcite that if the moving lattice stays aligned with the original applied longitudinal drive , a devil s staircase response occurs due to the matching of the symmetry directions of the lattice with the applied drive .
the width of a given step @xmath78 depends on the pinning strength and density . in fig .
[ fig : image](c ) we illustrate the evolution of the @xmath75 step for @xmath100 , 0.04 , 0.06 , and @xmath101 , showing the growth of @xmath78 with @xmath25 .
we also find that the step widths increase when the pinning density @xmath102 is increased . in fig .
[ fig : image](d ) we plot the evolution of @xmath78 as a function of pinning strength for @xmath103 , @xmath81 , @xmath75 @xmath82 , @xmath83 and @xmath84 .
we have verified that these results are robust up to systems of size @xmath104 provided that the lattice remains free of dislocations . in two - dimensional pinned systems it has been argued that dislocations eventually appear at large distances even for weak disorder .
these dislocations would limit the resolution of the devil s staircase , which should remain observable as long as the distance between dislocations is comparable to or greater than the system size .
vs @xmath70 for a moving vortex lattice with the same parameters as in fig .
1(a ) , but with @xmath105 .
( b ) the fraction of six - fold coordinated vortices @xmath106 vs @xmath70 for the same sample showing the correlations of the drops in @xmath106 with the transitions in @xmath74 .
, width=336 ] in the upper range of the intermediate pinning strength regime @xmath107 , the longitudinally driven vortices form a moving smectic state and the transverse depinning transition is _ plastic . _ here the transverse velocity response is characterized by a series of sharp disordering - reordering transitions , giving rise to consecutive pronounced steps in @xmath74 , as shown in fig .
[ fig : iv](a ) for a system with @xmath108 .
this ordering - disordering velocity - force response , which is distinct from a devil s staircase structure , occurs when the pinning is strong enough that the particles move in channels that are separated by a number of aligned dislocations , which is indicative of a smectic ordering .
we also find two prominent peaks in the structure factor @xmath109 consistent with a smectic state , as illustrated in fig .
[ fig : sk ] . here
in addition to the two prominent peaks , fig . [
fig : sk ] shows the presence of four smaller peaks .
since @xmath111 falls at the edge of the regime where the moving smectic state forms , there are only a small number of dislocations present in the moving smectic state and correlations between neighboring channels of vortices occur over a width of several channels .
these correlations produce the four smaller peaks in @xmath109 . for larger system sizes
@xmath112 these four small peaks shrink in magnitude while the two prominent peaks remain prominent . in ref .
@xcite , the prominent peaks were observed to decay algebraically with system size and the state was termed a moving transverse bragg glass .
it is beyond the scope of this work to confirm the existence of a moving transverse bragg glass . for larger values of @xmath25 , the dislocation density in the moving smectic state increases
, the correlation length between the moving vortex channels decreases to the width of a single channel , and only two peaks appear in @xmath109 . for @xmath113 the channel structure
is destroyed and the vortices move plastically . for a system with @xmath108 in the moving smectic state at @xmath114 and @xmath115 .
the central peak is not shown . ,
width=316 ] taken from a single snapshot .
small dots : sixfold - coordinated vortices ; large dots : nonsixfold - coordinated vortices .
( b , d , f ) : vortex trajectories ( black lines ) during a fixed time span .
( a , b ) : transversely pinned state at @xmath116 where a clear channel structure appears .
( c , d ) : the transition state at @xmath117 .
( e , f ) : the state at the first plateau in @xmath74 at @xmath118 where the system has reorganized into a tilted channel structure . , width=336 ] since the transverse depinning process at @xmath108 is plastic , a large number of transient dislocations are induced at the depinning transition which destroy the well ordered channel structures . in fig .
[ fig : iv](b ) we plot the fraction of sixfold - coordinated vortices @xmath106 for the same system in fig .
[ fig : iv](a ) , indicating that the changes in the behavior of @xmath74 coincide with changes in @xmath106 . in the regions where @xmath74 is increasing linearly with drive
, @xmath106 drops and shows rapid fluctuations . here
, the 1d channel structure of the moving lattice is destroyed .
we note that for @xmath119 we do not observe any dislocations in the longitudinal moving lattice ; however , since the system is close to the value of @xmath120 at which dislocations appear in the longitudinally moving lattice , at the transverse depinning transition the extra strain on the lattice is enough to induce some dislocations and the system also depins plastically . to further illustrate the change in lattice structure that occurs on the steps in the intermediate pinning regime , in fig
[ fig : delaunay](a ) we show the voronoi construction of the vortex lattice from the system in fig . [ fig : iv ] for a transversely pinned state at @xmath116 .
there are a small number of dislocations present , indicated by large dark circles , and all of the dislocation pairs are aligned with the direction of the drive ( to the right in the figure ) .
the vortex lattice is also aligned with the driving direction . in fig .
[ fig : delaunay](b ) the vortex positions ( black dots ) and trajectories ( black lines ) indicate that the motion is confined to well defined 1d channels . above the transverse depinning transition , illustrated in fig .
[ fig : delaunay](c , d ) for @xmath121 , numerous dislocations are present as indicated by fig .
[ fig : delaunay](c ) , while the trajectories in fig .
[ fig : delaunay](d ) show that the channeling effect has been lost .
this disordered type of flow appears both at the transitions among the higher steps as well as in the regions where @xmath74 is increasing linearly . in fig .
[ fig : iv](a ) , the velocity jumps to a finite value and becomes locked onto a plateau for @xmath122 .
here and on the other plateaus in @xmath74 , the lattice moves in a channel structure which is aligned at an angle to the @xmath33 direction , such as that shown in fig .
[ fig : delaunay](e , f ) for @xmath123 . as @xmath70 increases , the system undergoes repeated plastic depinning transitions , each of which appears as a drop in @xmath124 . in the plastic flow region that follows each of these depinning transitions , the vortices move in the direction of the net force vector . as the transverse drive increases to a new plateau , the lattice reorders into a new channel structure which sets up a new transverse barrier to further increases in @xmath74 .
the channels require some time to reform , so the width of the reordering transition as a function of transverse driving force is dependent on the sweep rate of @xmath79 . for higher sweep rates , the width of the disordered regions between plateaus grows .
in the adiabatic regime , the steps would be much sharper , suggesting that the transitions are first order in nature .
we also find hysteresis in the steps when we reverse @xmath70 .
for very strong pinning , @xmath113 , and fixed @xmath64 , the channel structures are destroyed and the vortices flow plastically for all transverse drives ; however , for higher values of @xmath58 , a partially ordered phase appears and the transverse depinning occurs in the same manner as in fig .
[ fig : iv ] . in general , for stronger disorder the longitudinal moving state contains numerous dislocations and the transverse depinning occurs in a step like fashion .
the transition between devil s staircase behavior at weak pinning and the order - disorder transitions associated with intermediate pinning strength is determined by the appearance of dislocations in the longitudinally driven lattice . in general , a devil s staircase appears when the disorder is weak enough that there are no dislocations present at any drive . for this reason , no evidence of a devil s staircase appeared in the simulations of ref .
@xcite , which showed the existence of a transverse depinning barrier .
these simulations were performed in the intermediate pinning limit , as indicated by the linear scaling of @xmath125 with the pinning strength in ref .
@xcite and the presence of aligned dislocations in the lattice .
the ordering - disordering steps that we find in the intermediate pinning limit were not observed in ref .
@xcite since the resolution of the transverse force versus velocity curves was too low .
the existence of a weak pinning regime may depend on system size .
it has been argued that for any 2d system with random disorder , dislocations will always appear at sufficiently long length scales @xmath126 even if the disorder is weak . in our simulations
we have considered numerous different system sizes , and for @xmath59 we only observe dislocation free lattices .
due to computational constraints we could only simulate systems of size @xmath127 and less .
it is possible that for much larger systems , dislocations will eventually appear .
if enough dislocations are present to destroy the orientational order , the devil s staircase behavior would be lost when the well defined lattice directions of the moving lattice are lost .
it is then possible that for system sizes @xmath128 the devil s staircase behavior will occur , while for @xmath129 , the strong pinning behavior will be observed . on the other hand ,
if the system is in 3d with point pinning but with stiff vortices , than a 3d moving bragg glass could form without dislocations , permitting a devil s staircase behavior to be observed for all lengths .
we also note that the argument that dislocations will appear at some length scale @xmath130 was made in the context of equilibrium systems . in the moving vortex case
it may still be possible to obtain a defect free lattice in two dimensions as suggested by koshelev and vinokur @xcite . versus @xmath120 for @xmath64 showing the three different response regimes .
for weak pinning , @xmath69 , the transverse depinning is elastic . for intermediate pinning , @xmath131 ,
the transverse depinning is plastic .
there is no transverse depinning barrier for strong pinning @xmath132 .
, width=336 ] in fig . 5 we plot the transverse depinning barrier @xmath133 vs @xmath25 and
indicate the three different pinning regimes which can be distinguished by the nature of the transverse response . in the weak pinning regime for @xmath134 ,
the transverse depinning is elastic and a devil s staircase forms when @xmath135 is swept . within this regime
, @xmath136 increases monotonically with @xmath120 . for the intermediate pinning regime @xmath137 ,
the transverse depinning is plastic and @xmath136 is much higher than it was in the elastic depinning regime . for strong pinning @xmath138 ,
the vortex lattice does not reorder when the longitudinal drive is applied , and the transverse depinning barrier disappears .
vs @xmath70 for a sliding wigner crystal with fixed @xmath139 .
( b ) @xmath106 vs @xmath70 .
( c ) @xmath140 vs @xmath70 .
( d ) the particle trajectories for the transversely pinned sliding state at @xmath141 .
( e ) the hysteresis in the @xmath74 vs @xmath70 curve .
lower curve : increasing @xmath79 .
upper curve : decreasing @xmath79 .
, width=336 ]
we now consider the effect of a smoothly varying potential substrate of the type that arises in wigner crystal systems for charge disorder . in general , with this type of disorder we observe much larger relative critical transverse depinning thresholds @xmath142 than with short range pinning . for
long range disorder @xmath143 can be as high as @xmath144 , while for short range disorder it is less than @xmath145 to @xmath146 .
the effective shaking temperature @xmath147 , which arises from the rapidly fluctuating force caused by the pinning , for particles moving over short range random disorder at velocity @xmath148 is @xmath149 @xcite . at high drives ,
the effective temperature drops with increasing @xmath148 , and the system can reorder . at lower drives , @xmath148 is smaller and the system melts into the plastic flow phase . in the smectic phase , the transverse depinning barrier increases as the longitudinal drive is decreased @xcite so that the maximum transverse barrier occurs just before the longitudinal drive is so low that the shaking temperature melts the lattice and @xmath70 is lost . for long range , smoothly disordered substrates , there is no rapidly fluctuating force so the concept of an effective temperature does not apply .
this means that a partially ordered moving state can persist to drives much closer to the longitudinal depinning threshold than in the case of short range pinning . in fig .
[ fig : phase](d ) , the particle trajectories for the case of long range charge disorder show that the channels are much more meandering than in fig .
[ fig : delaunay](b ) . in fig .
[ fig : phase](a , b , c ) we plot @xmath74 , @xmath106 , and the longitudinal velocity @xmath140 for the sliding wigner crystal system with fixed @xmath150 . in this case
the system initially forms a moving smectic system with @xmath151 .
a step structure appears in @xmath74 . on the step
, the particles are flowing in channels and @xmath106 shows only small fluctuations . in regions
where @xmath74 is linearly increasing , @xmath106 rapidly fluctuates around a lower value , indicating that the system is more disordered .
the order - disorder transitions also affect @xmath140 , which decreases and shows strong fluctuations in the disordered regimes . in the ordered regimes ,
there are few fluctuations in @xmath140 and there is a trend for @xmath140 to increase with @xmath79 .
this effect is similar to loading - unloading cycles of the type found in friction systems . in fig
[ fig : phase](e ) we illustrate the presence of hysteresis in @xmath74 vs @xmath70 , which gives evidence that the order - disorder transitions have a first order characteristic .
in summary , using numerical simulations we have confirmed the prediction of giamarchi and le doussal that a devil s staircase transverse response occurs for an elastic vortex lattice moving over random pinning when the net force vector aligns with the symmetry directions of the moving lattice
. the devil s staircase response appears in the weak pinning regime when both the equilibrium state and the moving state are free of dislocations and the transverse depinning occurs elastically without the generation of defects . for intermediate pinning strengths , where the equilibrium state is dislocated and the vortices form a moving smectic state under a longitudinal drive , the transverse depinning occurs in a series of hysteretic order - disorder transitions where the system alternately becomes disordered and then reorganizes into channel structures that exhibit a transverse depinning threshold . in this case
a series of pronounced steps in the transverse response appear . in the case where the quenched disorder is long - range and smoothly varying ,
the order - disorder transitions dominate the response to an increasing applied transverse force .
these results are applicable to vortices and wigner crystals moving over random disorder . | using numerical simulations we show that , in the presence of random quenched disorder , sliding superconducting vortices and wigner crystals pass through a variety of dynamical phases when an additional transverse driving force is applied .
if the disorder is weak , the driven particles form a moving lattice and the transverse response shows a devil s staircase structure as the net driving force vector locks with the symmetry directions of the moving lattice , in agreement with the predictions of le doussal and giamarchi . for strong disorder , and particularly for smoothly varying potential landscapes , the transverse response consists of a sequence of disordering transitions with an intervening formation of stable channel structures . |
droplet - based microfluidics is a promising tool for performing biochemical or chemical assays or to observe liquid - liquid extraction @xcite .
droplets are unit systems of controlled volume and content , within which mixing , reacting and/or transferring can be achieved . therefore , a comprehensive understanding of droplet migration in microchannels is essential to many microfluidics applications .
rectangular channels with a width significantly larger than the height are commonly used in microfluidics devices . in such cases ,
the confinement of the droplet is only in the vertical direction and the droplet is exposed to a two - dimensional poiseuille flow ( as in a hele
shaw cell ) .
the migration of the droplet is driven by the ambient fluid with an average inflow velocity @xmath0 and a constant terminal migration velocity @xmath1 will be reached .
the ratio between the horizontal radius of the droplet ( @xmath2 ) and half of the channel height ( @xmath3 ) has been shown to have a strong effect on the droplet migration velocity @xmath1 in a microchannel ( hereafter we simply refer to @xmath1 as the droplet velocity since the transient process is not of interest in the present study ) .
the overall variation of @xmath1 with @xmath4 can be seen in the experiments by roberts _
et al._@xcite , ( the experimental results will be shown later in section [ sec:3d ] ) and can be divided into three different regimes according to the dominant mechanism : * poiseuille - dominated regime ( regime p ) , @xmath5 . in this regime
the droplet velocity is dominated by the poiseuille flow in the hele shaw cell . in the limit of @xmath6 ( as the droplet is a point mass ) the droplet at the centerline of the channel will move with the maximum fluid velocity ( @xmath7 ) . as @xmath2 increases the droplet
is exposed to the parabolic fluid velocity profile with a decreasing velocity toward the top and bottom walls and the droplet slows down .
theoretical studies on the migration of droplets much smaller than the channel height have been conducted in many previous works , see for example nadim and stone @xcite and hudson @xcite .
* film - dominated regime ( regime f ) , @xmath8 : in this regime , the droplet overfills the channel and a thin lubrication film is formed between the droplet and the wall .
the droplet loses its roughly spherical geometry and becomes `` pancake - like '' .
furthermore , the droplet velocity @xmath1 becomes independent of @xmath4 and is mainly dictated by the lubrication film .
the film thickness @xmath9 is governed by the balance between surface tension and viscous stresses .
as the former tends to minimize the interface curvature and pushes the droplet interface closer to the wall , the latter tends to open up the gap between the droplet and the wall .
therefore the capillary number ( @xmath10 , where @xmath11 and @xmath12 is the viscosity of the ambient fluid and the surface tension , respectively ) , which is the ratio between the viscous and surface tension forces , is critical in determining the film thickness and the droplet mobility in this regime @xcite . beyond the capillary number , the viscosity ratio between the droplet and the ambient fluid has also been shown to have a significant influence on the lubrication film thickness and the flow pattern @xcite .
* transition regime ( regime t ) , @xmath13 : this regime connects the previous two regimes .
as the droplet radius is comparable to half of the channel height , the effect of confinement on the droplet becomes significant . as a result ,
the droplet velocity decreases dramatically with @xmath4 in this regime .
the three droplets shown in fig .
[ fig : problem_description ] schematically represent these three regimes .
the axes corresponding to the channel length , height , and width are defined as @xmath14 , @xmath15 , and @xmath16 , respectively .
the experimental and numerical results of roberts _ et al._@xcite clearly show the variation of droplet velocity in the poiseuille - dominated regime , and close agreement with theoretical predictions has been achieved .
the experimental data in the film - dominated regime is less documented however .
furthermore , simulation results for the transition and film - dominated regimes are not provided , due to the numerical challenge in resolving the thin film @xcite .
the present study aims at improving the current understanding of droplet motion in microchannels , in particular the droplet dynamics in the transition and film - dominated regimes .
in contrast with scenarios in which the droplet is much smaller than the channel height , the dynamics of droplets in the film - dominated regime ( _ i.e. _ , @xmath17 ) are more complicated and less understood .
early efforts in predicting the velocity of a flat bubble in a hele
shaw cell go back to the pioneering theoretical work by taylor and saffman @xcite .
shaw equations are obtained by averaging the stokes equations along the channel height ( depth ) , and it can be shown that the bubble velocity @xmath1 is always larger than the average inflow velocity @xmath0 . for low capillary numbers , the horizontal cross - section of the bubble is close to a circle ( exhibiting a pancake shape similar to the present problem ) @xcite , then @xmath18 according to the taylor
saffman theory @xcite . in the later experiments by kopf - sill and homsy @xcite
, the bubble velocity was observed to be about an order of magnitude lower than the taylor saffman prediction .
possible reasons for the discrepancy have been explored , such as an excessive drag due to a moving contact line within the film and marangoni effects due to surfactant transport @xcite , yet the underlying mechanisms responsible for the low mobility of the bubble is still not fully understood . in the limit when the 2d droplet is perfectly circular , the hele
shaw equations have also been solved analytically by afkhami and renardy @xcite and gallaire _
et al._@xcite to obtain the droplet velocity for arbitrary viscosity ratios as @xmath19 where @xmath20 is the viscosity of the droplet fluid . equation ( [ eq : hs_drop_vel ] ) recovers the taylor
saffman solution @xmath18 for bubbles ( @xmath21 ) .
nevertheless , since the hele
shaw equations ignore the lubrication film , eq .
( [ eq : hs_drop_vel ] ) does not capture the effect of the film dynamics on the droplet velocity .
due to the small scale of microchannels , experimental measurements of the detailed flow structure are very difficult .
recently huerre _
et al._@xcite presented quantitative optical measurements of lubrication film thicknesses as small as 20 nm for a channel height @xmath22 m
( for such a thin film , the intermolecular forces become important for the film dynamics and droplet velocity . in the present study
, we focus on the cases where the minimum film thickness is always larger than one micrometer . as a result ,
the effects of intermolecular forces can be ignored . )
however it is very challenging to probe local values of velocity and pressure , and consequently the flow structure inside and outside of the droplet is hard to obtain . numerical simulations which provide a complete description of the flow field
can shed light on the present problem .
it has been shown that numerical approximations of the navier - stokes equations can represent flow physics in many microfluidics applications , such as droplets / bubbles migration and the breakup of droplets at t - junctions @xcite .
nevertheless 3d direct numerical simulations of droplet migration in a microchannel are challenging as well .
this is in particular true for droplets with large ratio @xmath4 and small ca .
since capillary effects are dominant , the time step used in the temporal integration is controlled by the surface tension term in the momentum equation . for small capillary numbers
the time step becomes very small and thus the number of steps for the terminal velocity of droplet to be reached is huge .
in addition , at low capillary numbers the thickness of the lubrication film becomes very small ( several orders of magnitude smaller than the channel height ) @xcite .
since the mesh resolution and the time step restriction are dictated by the smallest length scale in the problem , the total number of cells and integration steps in 3d simulations are both very large .
three - dimensional simulation results of droplet / bubble in microchannels are rarely seen until the very recent works by hoang _
et al._@xcite , roberts _ et al._@xcite and zhu and gallaire @xcite .
however , as far as the authors are aware , a 3d direct numerical solution to the navier - stokes equations for droplets with a large @xmath4 and with a lubrication film formed has never been reported in the literature .
the goal of the present study is to investigate the droplet migration in a hele shaw cell through 3d direct numerical simulations .
the capillary number of the droplet is considered to be small enough so that the horizontal shape of the droplet is near - circular all the time .
particular attention will be paid to the effect of the lubrication film on the droplet dynamics .
a two - phase flow solver , gerris@xcite , is employed to solve the variable - density navier - stokes equation with surface tension . in the gerris solver ,
the volume - of - fluid ( vof ) method is used to resolve the interface between the two phases and an adaptive mesh is used to refine the mesh only in the lubrication film and near the interface , so that the computational cost can be reduced .
the governing equations and the numerical methods are presented in section [ sec : equations ] .
the simulation results will be shown in sections [ sec:2d ] and [ sec:3d ] .
we will first perform simulations of 2d droplets migrating in a microchannel with parameters similar to the 3d case for the purpose of validation , see [ sec:2d ] .
the simulation results and discussions for the 3d droplet are then presented in section [ sec:3d ] .
finally , we draw conclusions on the present study .
the one - fluid approach is employed to resolve the two - phase flow , where the phases corresponding to the droplet and the ambient fluid are treated as one fluid with material properties that change abruptly across the interface .
the incompressible , variable - density , navier - stokes equations with surface tension can be written as @xmath23 where @xmath24 , @xmath25 , @xmath26 , and @xmath27 represent density and viscosity , velocity and pressure , respectively .
the deformation tensor is denoted by @xmath28 with components @xmath29 .
the third term on the right hand side of eq .
( [ eq : mom ] ) is a singular term , with a dirac distribution function @xmath30 localized on the interface , and it represents the surface tension force .
the surface tension coefficient is @xmath12 , and @xmath31 and @xmath32 are the local curvature and unit normal of the interface .
the volume fraction @xmath33 is introduced to distinguish the different phases , in particular @xmath34 in the computational cells with only the ambient fluid ( respectively @xmath35 in the droplet ) , and its time evolution satisfies the advection equation @xmath36 the fluid density and viscosity are then defined by @xmath37 where the subscripts @xmath38 and @xmath39 represent the ambient fluid and the droplet respectively . the navier - stokes equations ( eqs .
( [ eq : mom ] ) and ( [ eq : divfree ] ) ) are solved by the open - source package gerris@xcite . in gerris , a finite volume approach based on a projection method is employed .
a staggered - in - time discretisation of the volume - fraction / density and pressure leads to a formally second - order accurate time discretisation .
the interface between the different fluids are tracked and followed using a volume - of - fluid ( vof ) method @xcite .
a quad / octree spatial discretisation is used , which gives a very important flexibility allowing dynamic grid refinement into user - defined regions @xcite .
finally the height - function ( hf ) method is utilized to calculate the local interface curvature , and a balanced - force surface tension discretization is used @xcite .
both temporal and spatial schemes of integration are explicit .
therefore , numerical stability requires the time step for integration to satisfy the stability conditions corresponding to the advection , viscous , and surface tension terms , respectively . due to the small capillary number in the present problem , the time step restriction is dominated by the contribution from the surface tension force .
the conventional capillary time step restriction is given by brackbill _
et al._@xcite as , @xmath40 where @xmath41 represents the cell size . for droplets with large @xmath4 ,
the thickness of the lubrication film is the smallest scale in the problem and thus dictates the smallest cell size and the time step to be used .
taking the conventional scaling of film thickness @xcite , the grid size can be estimated as @xmath42 .
( the scaling of the film thickness on capillary number will be discussed in detail in the sections [ sec:2d ] and [ sec:3d ] . ) then eq . (
[ eq : dt_surf_tension ] ) becomes @xmath43 if we assume that the terminal migration velocity is reached after a time of order @xmath44 , then the total number of time steps can be estimated as @xmath45 where @xmath46 .
as the total number of grid points ( on a regular cartesian grid ) will scale like @xmath47 with @xmath48 the dimension of the problem .
then the total work ( number of grid points times number of time steps ) for a 3d problem will thus scale as @xmath49 given that capillary numbers for microfluidics applications can be as low as @xmath50 , we see that direct numerical simulations will be extremely expensive .
the adaptive mesh refinement in gerris allows to refine the mesh only in the lubrification film , which greatly decreases this cost , however this is still not sufficient to reach such low capillary numbers .
a different capillary time step restriction has been proposed by galusinski and vigneaux @xcite .
they suggested that when both @xmath51 and @xmath52 are very small , the capillary time step should satisfy @xmath53 where @xmath54 is the capillary time step related to the stokes equation ( when inertial effects are negligible ) .
it can then easily be shown that the ratio @xmath55 . therefore , if @xmath56 the larger time step @xmath57 can be used . in the present study , however , we are interested in the cases where @xmath58 and @xmath52 varies from @xmath59 to @xmath60 ( see the detailed values in section [ sec:3d ] ) . as a consequence ,
the criterion @xmath56 is never satisfied and eqs .
( [ eq : dt_surf_tension])-([eq : totalwork_surf_tension_bkz ] ) remain valid to estimate the time step and computational cost .
before we present the 3d results , 2d simulations for a droplet moving in a microchannel are performed for validation purpose .
the simulation setup is schematically shown in fig .
[ fig : problem_description_2d ] .
the flow is symmetric in the @xmath15 direction with regard to the axis @xmath61 , thus only the upper half of the channel is considered .
the inflow from the left boundary is a parabolic velocity profile with mean value @xmath0 and maximum value @xmath7 .
an outflow boundary condition is applied at the right boundary of the domain ; while the top boundary is taken as a no - slip wall condition .
the length of the bubble is denoted by @xmath62 , and @xmath63 .
as the droplet is constrained by the channel , a lubrication film is formed between the droplet and the wall .
the channel length is @xmath64 , which is long enough for the droplet to reach its terminal velocity @xmath1 before being affected by the outflow boundary .
the physical properties of the ambient fluid and the droplet are listed in table [ tab : phys_parameters ] , which are similar to the recent experiments by roberts _
et al._@xcite the reynolds number based on the channel height and average inflow velocity , @xmath65 , is taken to be 1.35 .
the surface tension is varied from 0.0002 to 0.0159 n / m .
.[tab : phys_parameters ] physical properties of the ambient fluid and the droplet in the present study .
denotes dimensional parameters . ]
[ cols="^,^,^,^,^,^",options="header " , ] the physical parameters are identical to the 2d simulation presented in section [ sec:2d ] ( see table [ tab : phys_parameters ] ) .
parametric studies are performed by varying the surface tension coefficient @xmath12 and the droplet radius @xmath2 .
taking the ambient fluid density @xmath66 , the half of the channel height @xmath3 , and the mean inflow velocity @xmath0 as characteristic scales , the unknown variables and physical parameters can be non - dimensionalized and the governing equations ( eqs .
( [ eq : mom])([eq : color_func ] ) ) are solved in dimensionless form .
the key dimensionless parameters of the present problem are given in table [ tab : dmls_parameters ] .
as shown in the table , the ratio @xmath4 and the capillary number based on the average inflow velocity @xmath67 are chosen as the tuning parameters .
the capillary numbers in the 3d simulation are significantly larger than that in experiments by roberts
_ et al._@xcite , in which @xmath68 . the smallest @xmath67 considered here is about an order of magnitude larger than this experimental value .
as illustrated in the previous 2d results , the lubrication film formed is very small for small capillary numbers .
it is prohibitively difficult to simulate droplets with a larger @xmath69 and a small @xmath67 comparable to the experimental value with current computational resources .
nevertheless it is expected that as long as the thickness of the lubrication film does not reach the regime where van der waals forces are important , the effect of the lubrication film on the droplet dynamics can still be captured with simulation results at larger @xmath67 . at the end of this section an extensive discussion on the effect of surface tension will be given . ,
( c ) thickness ( height ) of the gap ( film ) between the droplet interface and the top channel wall , denoted by @xmath9 , and ( d ) relative @xmath70-velocities at the top and the side of the droplet ( @xmath71 and @xmath72 ) as a function of the ratio between the droplet radius @xmath2 and half of the channel height @xmath3 .
the experimental and theoretical results are shown for comparison @xcite . ]
the temporal evolution of the droplet velocity is shown in fig .
[ fig : udvsr](a ) for @xmath73 and @xmath4 varying from 1 to 2
. when @xmath4 increases , @xmath1 decreases and in contrast , the time it takes for the droplet to reach its terminal velocity increases ( while identical initial conditions are imposed ) . from @xmath74 to 1.25 , @xmath1 decreases significantly ( about 9% ) .
the decrease of @xmath1 with @xmath4 becomes more gradual for larger @xmath4 .
it is seen that the difference between @xmath1 for @xmath75 and @xmath63 becomes invisible as @xmath76 .
the simulation results of the droplet velocity as a function of @xmath4 are shown in fig .
[ fig : udvsr](b ) . the experimental and simulation results of roberts _ et al._@xcite and the theoretical results of nadim and stone @xcite for small @xmath4
are also shown for comparison .
the present predictions of @xmath1 are in good agreement with both the theoretical and experimental results for @xmath77 .
for @xmath78 the simulation results of @xmath1 are larger than those measured in the experiment .
the discrepancy is expected since smaller surface tension coefficients ( or larger capillary numbers ) are considered in the simulation .
nevertheless the simulation results capture well the trend of variation of @xmath1 with @xmath4 , and the three different regimes ( poiseuille - dominated , transition , and film - dominated regimes ) can be clearly seen .
in particular , the variation of @xmath1 in the transition and the film - dominated regimes is much better illustrated . in the transition regime ( @xmath79 )
, the droplet just overfills the channel forming a lubrication film .
correspondingly , the droplet velocity decreases rapidly . in the film - dominated regime , for @xmath80 , @xmath1
is observed to reach a plateau gradually .
the plateau value of @xmath1 decreases with @xmath67 , approaching the experimental value which corresponds to a much smaller @xmath67 .
the three different regimes can also be identified from the variation of film thickness with @xmath4 as shown in fig .
[ fig : udvsr](c ) . in the poiseuille - dominated regime ,
as the gap between the droplet and the top wall is relatively large , its thickness is not affected by the wall .
the droplet remains close to a sphere , so @xmath81 closely follows @xmath82 . as @xmath4 approaches 1 , ( in the transition regime , ) the film thickness starts to deviate from @xmath82 and becomes governed by the viscous stress and the surface tension .
when @xmath8 , ( in the film - dominated regime , ) the film thickness becomes completely independent from the horizontal radius and becomes dependent on @xmath67 . at last the droplet circulation velocities in the three different regimes are shown in fig .
[ fig : udvsr](d ) .
the @xmath70-velocities at the top ( the point with maximum @xmath15 on the droplet ) and the side of the droplet ( the point with minimum @xmath16 on the droplet ) are denoted as @xmath71 and @xmath72 , respectively .
the simulation results of roberts _ et al._@xcite and the theoretical results of hudson @xcite are shown for comparison .
for @xmath83 , namely in the poiseuille - dominated regime , @xmath71 and @xmath72 vary as parabolic functions with respect to @xmath4 .
the present predictions of @xmath71 are in good agreement with both the previous numerical results and the theoretical prediction . in the film - dominated regime ( @xmath84 )
, it is observed that both @xmath71 and @xmath72 reach constant values , similar to @xmath1 .
it is also interesting to notice that the relative velocities @xmath85 and @xmath86 for different @xmath67 seem to overlap .
an interesting observation from the simulation results is that the velocity of a 3d droplet can vary in a much wider range than the velocity of a 2d droplet .
in particular , the droplet velocity in the film - dominated regime is observed to be generally smaller than the average inflow velocity .
this low droplet velocity has also been observed in experiments @xcite . from fig .
[ fig : udvsr](b ) it can be seen that when @xmath4 increases from 0.5 to 2 , the droplet velocity @xmath1 for @xmath87 decreases from 1.4 to 0.8 . in this section ,
we propose a simple model of mass fluxes balance to qualitatively illustrate the droplet mobility in the film - dominated regime and the resulting 3d flow patterns .
[ sec : model ] in the 2d planar ( or axisymmetric ) geometry the ambient fluid which enters the channel can only move along the flow direction , and thus must push the droplet to move faster than the average inflow velocity to ensure mass conservation .
if @xmath9 is given the corresponding value of @xmath1 can be obtained by the balance of mass fluxes @xcite . taking a control volume ( of zero thickness ) near the rear of the droplet as shown in fig .
[ fig : mass_cons](a ) , the mass fluxes are identical to zero on both the @xmath15 and @xmath16 boundaries and thus the fluxes from the left and right must cancel .
considering a bubble with large @xmath4 and small @xmath81 , with a stress - free interface ( and thus a constant velocity within the film ) , the relation between @xmath9 and @xmath1 can be expressed as @xmath88 it is seen that @xmath89 for non - zero @xmath9 . for small @xmath81
the excess of droplet velocity is small and the droplet velocity @xmath1 is close to the average inflow velocity @xmath0 . in the 3d analogue , since the ambient fluid can flow around the droplet from the sides ( or edge ) of the droplet , the mass fluxes balance and the flow pattern becomes more complex .
similarly to the 2d case we look at a thin layer of small thickness @xmath90 along the symmetric plane ( @xmath91 ) near the rear of the droplet . as there is zero flux on the symmetric plane @xmath92 ,
the transverse mass flux is only non - zero at the plane @xmath93 , normalized by the flux from the right , _
i.e. _ , @xmath94 , which can be expressed as @xmath95 assuming the variation of @xmath0 in @xmath91 is negligible .
the transverse flux @xmath96 is not uniformly distributed on the plane @xmath93 , instead it is concentrated near the stagnation point at the rear of the droplet .
a similar analysis can be performed for the front of the droplet and it can be shown that @xmath96 and @xmath97 have the same magnitude but different signs , which creates a circulation around the edge of the droplet . for a given @xmath9 , @xmath96 is related to @xmath1 through mass conservation , however the fluxes balance is not sufficient to determine @xmath1 .
this extra spatial degree of freedom of mass flux thus allows a much wider variation of droplet velocity .
in contrast to the 2d case , where @xmath1 is always slightly larger than @xmath0 , in the 3d case @xmath98 has been observed to vary from as low as @xmath99 to about 2 in experiments @xcite .
furthermore the two scenarios of @xmath100 and @xmath101 correspond to two qualitatively different three - dimensional flow patterns over the droplet . assuming @xmath102 , it can be shown that @xmath103 when @xmath100 .
the pure " bubble experiment by park _
et al._@xcite falls in this scenario of high - migration - velocity ( hmv ) . in the droplet reference
frame the sweeping motion of the ambient fluid on the edge of the droplet is from the front to the rear .
combined with the fact that the flow on the top of the droplet is always from the front to the rear , this means that the overall flow in the droplet reference frame is in the rear - to - front direction . on the other hand the scenario of low - migration - velocity ( lmv ) ,
_ i.e. _ , @xmath101 and thus @xmath104 , is very different .
the droplets in the present simulation and in the experiments of roberts _ et al._@xcite and huerre _ et al._@xcite fall in this category .
the positive transverse flux near the droplet rear ( @xmath105 ) indicates that the fluid actually goes around the droplet from the rear to the front in the droplet reference frame . as a consequence ,
the relative flows on the top and the side of the droplets are in opposite directions , ( which are also illustrated in fig .
[ fig : udvsr](d ) ) .
a complex interfacial flow sweeping over the droplet is then formed , in which the flow ( in the droplet reference frame ) first goes from the front to the rear on the top of the droplet then returns from the rear back to the front on the edge of the droplet .
( a visualization of this interfacial flow will be presented in fig .
[ fig : interfacial_vel - vector_dropframe ] . ) and @xmath106 , 1 and 2 .
the color of the streamlines denotes the corresponding relative velocity magnitude .
the pink line indicates the droplet interface . ] and @xmath63 in the droplet reference frame : ( a ) internal circulation and external circulation near the front of the droplet .
( b ) flow within the film between the droplet and the wall and the internal circulation inside the droplet . the color of the streamlines denotes the magnitude of the relative velocity .
the angular coordinate @xmath107 is defined in the droplet reference frame with the origin located at the center of the horizontal cross - section . ] the three - dimensional flow structure inside and outside of the droplet is shown in fig . [
fig : streamlines ] for @xmath73 and @xmath106 , 1 and 2 .
the streamlines on the two symmetric planes @xmath92 and @xmath61 are plotted in the droplet reference frame and colored by the relative velocity magnitude . as the droplet moves faster for smaller droplets the relative velocity magnitude is larger for smaller @xmath4 .
the flow pattern for @xmath77 here is also consistent with the simulation results of roberts _ et al._@xcite .
it is interesting to compare the 3d flow pattern for the droplet with @xmath63 to its 2d analogue shown in fig .
[ fig:2d_results ] . in the vertical plane
@xmath92 the ambient fluid is moving from right to left at the top ( @xmath108 ) and in a reverse direction at the center plane ( @xmath61 ) . as a result two
external circulations develop outside of the droplet , similar to the 2d analogue shown in fig .
[ fig:2d_results](a ) .
the external flow will also drive internal circulation inside the droplet . similarly to the 2d case , six stagnation points ( a , b , c , d on the droplet interface and e , f inside the droplet ) can be seen , however the stagnation flow around these points becomes three dimensional .
for example in the 2d case , the flow reaches the rear stagnation point c and can only turn toward point a and then turn again to the left and leave the domain , forming a counter - clockwise external circulation , see fig .
[ fig:2d_results](a ) .
in 3d the fluid has an extra degree of freedom and can go in the @xmath16 direction around the droplet toward the stagnation point d , see the streamlines on the @xmath14-@xmath16 plane in fig .
[ fig : streamlines](c ) .
this transverse fluid motion near the stagnation point c is characterized by @xmath109 in the model presented in section [ sec : model ] . due to the 3d flow structure
the minor internal circulations ( a - e - c - a and b - d - f - b ) in the 2d case are invisible here .
the transverse fluid motion significantly affects the external and internal circulating flow . as shown in fig .
[ fig : streamlines_3d](a ) ,
the 3d streamlines inside and outside of the droplet become very complicated and are very dependent on the angular coordinate @xmath107 .
( here @xmath107 is defined as the angle between the normal to the interface in the @xmath14-@xmath16 plane and the @xmath14 axis . ) for @xmath110 the circulation is perfectly aligned with the symmetric @xmath14-@xmath15 plane , and the flow pattern is similar to the 2d case .
as @xmath107 increases the external circulation is twisted in a complex manner .
the incoming flows for nonzero @xmath107 near the top of the channel are observed to turn toward the center plane @xmath92 , ( indicated by black arrows in fig .
[ fig : streamlines_3d](a ) ) . as the flow approach the droplet interface
, it turns to the side to adjust to the droplet shape , ( see the magenta arrow ) . at about @xmath111 degrees
the streamlines outside of the droplet are almost tangential to the droplet interface .
therefore , in contrast with the typical quasi-2d flows in a hele shaw cell , droplet migration in the microchannel generates a fully 3d flow . on the droplet interface for @xmath73 and @xmath106 , 1 and 2 .
the color denotes the magnitude of the relative velocity . ]
the vectors of the interfacial velocity in the droplet reference frame are shown in fig .
[ fig : interfacial_vel - vector_dropframe ] .
the color of the interface indicates the relative velocity magnitude .
regions with white color indicate zero velocity , _
i.e. _ , stagnation regions on the droplet interface .
the four stagnation points ( a , b , c , d ) on the droplet interface and 3d stagnation flows in the vicinity can be clearly seen . driven by the flow within the film , the interfacial flow on the top of the droplet moves to the left ; while the flow on the edge near the center plane @xmath61 moves in the opposite direction . an interesting and complex interfacial flow is then formed on the droplet interface , which is highlighted with yellow dashed lines in fig .
[ fig : interfacial_vel - vector_dropframe ] . at
first the flow starts from the vicinity of stagnation point b ( indicated by the blue dashed line @xmath112@xmath113 ) , and goes toward the rear .
it is seen that the velocity vectors on the top interface are not parallel to the @xmath14 axis , instead they turn toward the side of the droplet . the angle between the top interfacial flow and the @xmath114 axis
is denoted by @xmath115 here , and it can be seen that @xmath115 varies significantly with the angle @xmath107 of the relative flow near the wall ( in the droplet reference frame ) . for @xmath110 , @xmath116 , but then @xmath115 increases significantly with @xmath107 up to about 45 degrees . then top interfacial flow is observed to curve back toward the center plane ( @xmath92 ) before reaching stagnation point a. after reaching point a the flow reverses its direction and moves toward the front stagnation point d along the edge of the droplet . for the top interfacial flow leaving from @xmath112@xmath113 with a large @xmath107 ,
the flow may not get a chance to turn back to stagnation point a before returning to the front . as a result
a ring of zero interfacial velocity ( the region in white color ) is formed on the edge of the top where the interfacial flow reverses direction . near
the rear of the droplet the expected up - going interfacial flow from stagnation points c to a is not seen due to the strong interfacial flow along the edge .
the stagnation flows near c and a seem to merge , forming a stagnation area at the droplet rear .
furthermore , the streamlines in fig .
[ fig : streamlines_3d](b ) show that the flow within the lubrication film is influenced by the complex interfacial flow on the droplet interface .
similar to the interfacial flow on the top of droplet ( see fig . [
fig : interfacial_vel - vector_dropframe ] ) , when the flow passes through the lubrication film , the streamlines off the center plane @xmath92 also first turn toward the side and then return back to the streamwise direction . and ( b ) shear @xmath117 on the interface for @xmath73 and @xmath106 , 1 and 2 . the film for @xmath63 is highlighted by pink dotted lines .
] as the film between the droplet and the wall serves to lubricate " the droplet , when the film thickness decreases ( for example due to increasing surface tension ) it is expected the droplet will slow down correspondingly .
the increase of droplet velocity with the film thickness has been confirmed in the experiments by huerre _
et al._@xcite . to better understand the effect of the lubrication film on the droplet dynamics
, we plot the thickness of the lubrication film and the shear on the droplet top interface in fig .
[ fig : interface_height_shear ] for @xmath73 and @xmath106 , 1 and 2 .
as @xmath4 is small the gap between the droplet and the wall is large and thus strictly speaking there exists no film , the term is retained only for convenience . when @xmath4 is larger or comparable to unity a thin lubrication film with small thickness can be clearly seen . in particular for @xmath63 the film area ( highlighted by the pink dotted line ) becomes quite large compared to the cross - section of the droplet at the @xmath61 plane . instead of being centered , the lubrication film is shifted toward the rear ( see fig . [
fig : interfacial_vel - vector_dropframe ] ) . the overall film exhibits a
peach " shape .
values of the shear @xmath117 on the droplet interface are shown in fig .
[ fig : interface_height_shear](b ) . when @xmath4 is small the droplet is away from the top wall and the shear on the top of the droplet is indeed quite small . the shear increases dramatically as @xmath4 becomes comparable or larger than unity .
for @xmath63 it is observed that the shear on the droplet is not uniformly distributed and the large negative shear is located near the rear .
the region of high shear has a croissant shape , and minimum shear rate reaches about @xmath118 ( in dimensionless unit ) , the magnitude of which is more than twice larger than the minimum shear in the ambient 2d poiseuille flow , _
it is obvious that the strong shear stress induced by the film is closely related to the low mobility of droplets with larger @xmath4 . as mentioned at the beginning of this section , the capillary numbers considered in the present 3d simulations are larger than that in experiments . to justify that the simulation results presented above at least qualitatively reflect the correct flow physics of droplet migration in a hele
shaw cell , the capillary number is varied to investigate its effect on the results . from the droplet velocity previously shown in fig .
[ fig : udvsr](b ) , the change in @xmath1 is invisibly small for @xmath106 as @xmath67 varies from @xmath120 to @xmath121 .
though smaller capillary numbers are used , the surface tension is still sufficient to keep the droplet spherical and the droplet velocity is then mainly determined by the viscous stress .
the droplet velocity obtained in simulations for @xmath73 , which is two orders of magnitude larger than that in experiments , still agrees quite well with the experimental and theoretical results . on the other hand , as @xmath69 , the effect of
the capillary number on the droplet velocity becomes much stronger .
as shown in fig .
[ fig : udvsr](b ) and ( c ) , the droplet velocity and film thickness decrease by about @xmath122 and @xmath123 for @xmath63 , when @xmath52 decreases from @xmath120 to @xmath124 . the film thickness and shear on the droplet interface for @xmath87 and @xmath106 , 1 and 2 are shown in fig .
[ fig : interface_height_shear_smallst ] .
the surface tension here is four times that in fig .
[ fig : interface_height_shear ] while the other parameters remain the same .
( correspondingly @xmath67 in fig .
[ fig : interface_height_shear_smallst ] is a quarter of that in fig .
[ fig : interface_height_shear ] . )
when @xmath106 there is no difference in the gap thickness . for @xmath63 the film becomes significantly thinner as surface tension increases .
( therefore different legends for film thickness are used in in figs .
[ fig : interface_height_shear](a ) and [ fig : interface_height_shear_smallst](a ) for better visualization . )
the minimum film thickness for @xmath87 is about @xmath125 while for @xmath73 , @xmath126 .
it is also observed that when @xmath67 decreases , the lubrication film still shifts to the rear , however the shift becomes smaller and the droplet shape becomes more symmetric with regard to the @xmath15-@xmath16 plane .
furthermore , the overall peach " shape of the film remains , but the smallest film thickness for @xmath63 seems to move to the two sides of the film .
when @xmath67 decreases the general distribution of the shear rate within the film is similar .
the high - shear region still holds a croissant shape .
the minimum shear near the rear decreases to about @xmath127 which is about 25% larger than that for @xmath73 in magnitude .
this indicates that the decreases of @xmath67 results in a decrease of @xmath9 which in turn introduces a stronger shear drag on the droplet and reduces the droplet migration velocity . and ( b ) shear @xmath117 on the interface for @xmath87 and @xmath106 , 1 and 2 . ]
the minimal film thickness on the symmetric @xmath14-@xmath15 plane is plotted as a function of @xmath52 in fig .
[ fig : ud - h_vs_ca](a ) , where @xmath10 is the capillary number based on the droplet velocity @xmath1 . the 2d theory , _
i.e. _ , eq . ( [ eq : scaling_2d_inviscid ] ) , is also plotted for comparison .
it is seen that @xmath128 in the 3d case is much smaller than in the 2d case . for @xmath129 , @xmath128 in 3d
is an order of magnitude smaller than in 2d , as already visualized in figs .
[ fig:2d_results ] and [ fig : streamlines ] .
the difference between 3d and 2d film thickness seems to decrease with @xmath52 .
the film thickness variation with @xmath52 seems to follow a power law but the power coefficient is about 0.4 instead of 2/3 as in 2d .
the droplet velocity @xmath1 decreases with @xmath67 as shown in fig .
[ fig : ud - h_vs_ca](b ) . here
the capillary number based on the average inflow velocity is used following previous works @xcite . the droplet velocity obtained by the hele
shaw equations ( eq . ( [ eq : hs_drop_vel ] ) ) , which is about 0.85 , is also plotted for comparison .
the depth - averaged approximation misses the effect of the lubrication film on the droplet dynamics , and thus eq .
( [ eq : hs_drop_vel ] ) fails to capture the variation of the droplet velocity with capillary number .
it is seen that the hele
shaw solution under - predicts @xmath98 for larger @xmath67 and over - predicts @xmath98 for small @xmath67 .
the droplet velocity only matches the hele
shaw solution at @xmath130 .
similar observations have also been made by zhu and gallaire @xcite . the range of @xmath67 considered in the present simulation is an order of magnitude larger than the experimental value .
it is hard to tell if the simulation results would match the experimental measurements @xcite if the same @xmath67 is used .
nevertheless it seems like the computed @xmath1 would approach the experimental value as @xmath67 decreases . as discussed
previously the mass fluxes balance in 3d is substantially different from 2d . in the 2d case
it is known that @xmath131 ( for @xmath132 ) . in 3d
, @xmath133 becomes negative .
it is shown that @xmath1 also decreases with @xmath81 , but the decrease is non - linear and follows a power law about @xmath134 as indicated in fig .
[ fig : ud - h_vs_ca](c ) .
the simple model of transverse flux near the rear of the droplet ( eq . ( [ eq:3d_mass_cons ] ) ) can serve as a measure of the three - dimensional `` strength '' of the flow .
the variation of @xmath96 is plotted as a function of the film thickness in fig .
[ fig : ud - h_vs_ca](d ) .
when @xmath9 decreases it is observed that @xmath96 increases rapidly .
as the film becomes thinner , the drag induced by the lubrication film becomes stronger and it is more difficult for the ambient fluid to push the droplet forward . as a result the ambient fluid entering
the channel tends to go around the droplet from the transverse edge instead of pushing the droplet .
therefore it is quite clear that both the 3d flow effect and the lubrication film dynamics control the mobility of a droplet in a hele shaw cell .
plane and ( b ) droplet velocity as functions of the capillary number .
the droplet velocity obtained by the hele
shaw equations ( eq . ( [ eq : hs_drop_vel ] ) ) is also plotted for comparison .
( c ) excess of droplet velocity compared to the average inflow velocity and ( d ) normalized transverse flux near the rear of the droplet as a function of the film thickness .
] the present simulations are performed on 8 to 24 processors for 3 to 25 days depending on cases .
the adaptive meshes used for @xmath73 and @xmath124 and @xmath63 at the plane @xmath92 are shown in fig .
[ fig:2d - interface_with - mesh ] .
the main constraint on the computational efficiency comes from the small time step which is in turn restricted by the surface tension ( or capillary number ) , see eq .
( [ eq : dt_surf_tension ] ) .
when surface tension increases four times , the time step is divided by two and namely the computational time doubles assuming the mesh is unchanged .
however , as indicated in fig .
[ fig:2d - interface_with - mesh](b ) that if @xmath67 is further decreased , the minimum mesh size in the adaptive mesh , _
i.e. _ , @xmath135 , will not be sufficient and thus must be refined .
therefore , even with an adaptive mesh , it is too expensive to simulate droplets at capillary number comparable to experiments for large @xmath4 . to perform 3d direct simulation of droplets at low capillary number
, novel numerical schemes to compute the surface tension are required . if the surface tension term in eq .
( [ eq : mom ] ) can be treated in an implicit manner similar to what is done for the viscous term in stokes flow ( at low reynolds number ) , then the time - step restriction due to the explicit integration of the surface tension term can be lifted ( the authors are not aware of this type of scheme in the literature ) .
nevertheless the development of an implicit treatment of surface tension is out of the scope of the present paper and will be relegated to future work . for @xmath63 and ( a ) @xmath73 and ( b ) @xmath87 .
the migration of a droplet in a hele
shaw cell has been investigated by 3d direct numerical simulations .
we focused on the regime at low capillary number where the droplet remains circular in the horizontal plane .
parametric studies were performed by varying the droplet horizontal radius and the capillary number . for droplets with an horizontal radius @xmath2 larger than the half - height of the channel @xmath3 , the droplet overfills the channel and a thin lubrication film is formed between the droplet and the wall .
an adaptive two - phase flow solver is utilized for the simulations and mesh is locally refined to accurately resolve the thin film .
the ratio @xmath4 has a significant impact on the droplet velocity and controls three regimes : the poiseuille - dominated regime , the film - dominated regime , and the transition regime .
the droplet velocity decreases dramatically with @xmath4 in the transition regime . in the film - dominated regime ,
the droplet horizontal radius is much larger than the channel height and the droplet holds a pancake - like shape . then the migration velocity is found to become independent of the radius and is mainly dictated by the capillary number .
the simulation also shows that the droplet velocity in the film - dominated regime decreases with the capillary number and is generally lower than the average inflow velocity .
the lubrication film dynamics and the three - dimensional flow structure seem to both contribute to the low mobility of the droplet . in this low - migration - velocity scenario the interfacial flow on the top of the droplet moves from front toward the rear in the droplet reference frame , but reverses its direction moving toward the front from the two sides of the the droplet . as the depth - averaged hele
shaw equations ignore the effect of the lubrication film on the droplet dynamics , their solution fails to capture the dependence of droplet velocity on capillary number . as the shear induced by the lubrication film
has a strong influence on the droplet dynamics , the viscosity ratio can be an important parameter for the present problem , and we intend to investigate its influence in future work .
this project has been supported by the the anr tram project ( anr-13-bs09 - 0011 ) .
the simulations of this paper are conducted on our laboratory cluster and on the cines occigen machine for which we gratefully acknowledge grant x20152b7325 from genci .
32ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @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 * * , ( ) | droplet migration in a hele
shaw cell is a fundamental multiphase flow problem which is crucial for many microfluidics applications .
we focus on the regime at low capillary number and three - dimensional direct numerical simulations are performed to investigate the problem . in order to reduce the computational cost , an adaptive mesh is employed and high mesh resolution is only used near the interface .
parametric studies are performed on the droplet horizontal radius and the capillary number . for droplets with an horizontal radius larger than half the channel height
the droplet overfills the channel and exhibits a pancake shape .
a lubrication film is formed between the droplet and the wall and particular attention is paid to the effect of the lubrication film on the droplet velocity .
the computed velocity of the pancake droplet is shown to be lower than the average inflow velocity , which is in agreement with experimental measurements .
the numerical results show that both the strong shear induced by the lubrication film and the three - dimensional flow structure contribute to the low mobility of the droplet . in this low - migration - velocity scenario
the interfacial flow in the droplet reference frame moves toward the rear on the top and reverses direction moving to the front from the two side edges .
the velocity of the pancake droplet and the thickness of the lubrication film are observed to decrease with capillary number .
the droplet velocity and its dependence on capillary number can not be captured by the classic hele
shaw equations , since the depth - averaged approximation neglects the effect of the lubrication film . |
a multi - model cantor set is a metric space which has the following property .
there is a partition of @xmath0 into finitely many clopens @xmath1 ( called models ) so that given any point @xmath2 in @xmath0 and @xmath3 there is a neighborhood @xmath4 of @xmath2 with @xmath5 and a metric similarity mapping @xmath4 onto one of the models .
such a cantor set is described by a map @xmath6 which is a piecewise metric similarity ( an expanding map ) .
one may consider the cantor set as determined by the dynamical system @xmath7 .
the middle third cantor set is an example of a multi - model cantor set where one model suffices .
the middle third cantor set is self - similar . at every scale it contains identical copies of itself . in a sense
a multi - model cantor set has finitely many local pictures which are replicated , at different scales , everywhere .
this behaviour is similar to certain fractals , such as julia sets , which have a compact family of local geometries , ( rather than a finite family ) .
we associate to @xmath8 and a constant @xmath9 , an @xmath10 matrix @xmath11 with non - negative entries . roughly speaking
the @xmath12 entry of @xmath11 is the sum of the @xmath13-powers of the inverse scale factors of the similarities of the clopens ( later called clones ) of type @xmath14 contained in model @xmath15 .
this matrix determines the hausdorff measure and dimension as follows .
we use @xmath16 to denote the @xmath13-dimensional hausdorff measure . + * theorem 1 . * _ suppose @xmath9 and @xmath0 is a multi - model cantor set with matrix @xmath11 and @xmath17 is the frobenius eigenvalue for each @xmath11 .
let @xmath13 be chosen such that @xmath18 .
then the hausdorff measure of @xmath0 is finite and non - zero in this dimension .
therefore @xmath13 is the hausdorff dimension of @xmath0 .
also , let @xmath19 be the frobenius eigenvector of @xmath11 such that @xmath20 then @xmath21 . _ +
this result is given at the end of section [ hausdorff ] .
then in section [ linearity ] we investigate bilipschitz maps between multi - model cantor sets .
we prove that every bilipschitz map from one multi - model cantor set to another is measure linear ( has constant radon - nikodym derivative ) on some clopen .
this is a generalization of the results of cooper @xcite , pignataro @xcite and vuong @xcite , @xcite to this wider class of cantor set .
we use this in corollary [ 10.7 ] to provide an invariant up to bilipschitz homeomorphism of such cantor sets .
these results are from the author s master s thesis @xcite
we are concerned , in this paper , with the study of cantor sets with particular metrics . in the following sections we shall show some results about the hausdorff measure of these cantor sets . in order to make these results clearer to the reader we include the definitions of similarity map , @xmath22-bilipschitz map , and diameter and state two elementary results concerning hausdorff dimension
. + * definition . *
a surjective map @xmath23 between metric spaces is a _ similarity map _ if there is a constant @xmath24 such that for all @xmath25 in @xmath26 @xmath27 * definition .
* a map @xmath23 between metric spaces is _
@xmath22-bilipschitz _ if there is a constant @xmath24 such that for all @xmath25 in @xmath26 @xmath28 * definition . * for a set @xmath4 contained in a metric space @xmath26 , the _ diameter _ of @xmath4 is given by @xmath29 [ a ] suppose @xmath26 is a metric space and @xmath30 and the @xmath13-dimensional hausdorff measure of @xmath26 satisfies @xmath31 .
then @xmath13 is the hausdorff dimension of @xmath26 .
this follows directly from the definition .
[ multiply ] let @xmath26 , @xmath32 be metric spaces .
let @xmath23 be a surjective similarity map with scale factor @xmath24 , then @xmath33 this follows directly from the definition .
recall that a topological space @xmath0 is a _
cantor set _ if it is compact , totally disconnected , perfect and non - empty .
it is a fact that any two cantor sets are homeomorphic .
the situation becomes more interesting when one puts a metric on a cantor set and studies bilipschitz homeomorphisms , as shall be seen in section [ linearity ] .
in this paper we are interested in considering cantor sets which arise from dynamical systems .
we illustrate this by exhibiting a dynamical system which generates the middle third cantor set .
this dynamical system is determined by a _ clone structure _ which we shall now describe .
let @xmath34 $ ] , @xmath35 $ ] and @xmath36 $ ] and define the _ clone maps _
@xmath37 by @xmath38 then this _ clone structure _ is either denoted as @xmath39 or , because the domain @xmath40 is implicit in the definition of @xmath41 , it can also be denoted @xmath42 . to construct the middle third cantor set from the dynamical system we define
the map @xmath43 by @xmath44 observe that @xmath7 maps each of the @xmath40 linearly onto @xmath45 .
then @xmath46 in general @xmath47 is a disjoint union of @xmath48 intervals , each of length @xmath49 .
the middle third cantor set is @xmath50 we must note though that the clone structure is not unique . at each level
any clone can be replaced by a finite collection of clones which partition it .
therefore we usually refer to a cantor set @xmath0 , rather than @xmath51 . in figures [ middlethird3 ] and [ middlethird4 ]
we show two different clone structures for the middle third cantor set .
middle third cantor set with clone structure @xmath42.,width=396,height=75 ] middle third cantor set with clone structure @xmath52.,width=396,height=75 ] having chosen a clone structure we shall call a clopen a _ clone _ if it is the domain of a composition of clone maps in the clone structure . the empty composition is the identity map so the entire cantor set @xmath0 is also a clone .
the _ level _ of the cantor set is defined to be 0 .
the _ level _ of any other clone is defined to be @xmath53 if the smallest clone which properly contains it has level @xmath54 .
hence the middle third cantor set has two level-1 clones : @xmath55 $ ] and @xmath56.$ ] note that we will abuse notation and use @xmath57 to represent both the interval @xmath58 $ ] and also the subset of the cantor set contained in @xmath58 $ ] .
a clone @xmath59 is said to be at _ relative level 1 _ to a clone @xmath60 if given the level of clone @xmath60 as @xmath53 then the level of clone @xmath59 is @xmath61 .
the middle third cantor set is generated by 2 similarity maps , each with a similarity factor of 3 .
cantor sets can be generated by two or more maps with various similarity factors .
they can also be generated in higher dimensions in euclidean space @xcite , @xcite or in a general metric space @xcite .
below is a pictorial representation of a cantor set in @xmath62 .
this cantor set is given by a dynamical system ; a clone structure .
cantor set in @xmath62 with clone structure @xmath63.,height=170 ] figure [ 2dcantor5 ] is a cantor set in @xmath62 .
it has 3 level-1 clones , @xmath64 . below we describe how each level-1 clone is mapped onto the largest rectangle .
each clone contains a scaled down copy of the entire cantor set .
a clone structure on this cantor set can be described by three similarity maps .
if the origin is considered to be the top left vertex of the rectangle @xmath65 , then the map @xmath66 is a stretch by a factor of 2 , @xmath67 is a rotation by @xmath68 radians clockwise , a stretch by a factor of 2 and a translation and @xmath69 is a rotation by @xmath70 radians , a stretch by a factor of 4 and a translation .
in section [ middle third ] we discussed clone cantor sets in metric spaces arising from dynamical systems , where each clone can be mapped onto the whole cantor set by using a similarity clone map . in this section
we shall be looking at a new class of examples , called _ multi - model cantor sets _ which also arise from dynamical systems . in vuong
s thesis @xcite and paper @xcite , he studied clone cantor sets which can be determined by collections @xmath71 where the @xmath40 are level-1 clones of @xmath0 and the @xmath72 are level-1 clone maps of @xmath0 .
vuong was considering cantor sets where each clone is mapped onto the unit cube in @xmath73 by a clone map .
in other words he was considering cantor sets which look the same everywhere . in a sense a multi - model cantor set has finitely many local pictures which are replicated , at different scales , everywhere .
+ * definition .
* let @xmath74 be a metric space .
let @xmath1 be a partition of @xmath0 by clopens .
the @xmath75 are called _
models_. let @xmath76 be another partition of @xmath0 by clopens such that each @xmath77 is contained in some @xmath75 .
the @xmath77 are called _
level-1 clones_. we require that each @xmath75 contains at least 2 of the @xmath77 hence @xmath78 . for each @xmath77
there is a surjective similarity map @xmath79 for some @xmath15 .
we say that @xmath80 is a _ clone structure _ of a _ multi - model cantor set _ , @xmath0 .
as before we define the piecewise metric similarity @xmath81 + we may construct examples in euclidean space by using similarities of @xmath73 .
for example if @xmath65 is a hypercube in @xmath73 we obtain a nested set of clopens @xmath82 where @xmath83 then the multi - model cantor set is the intersection @xmath84 each model and each clone in a multi - model cantor set has a
_ type_. by definition each model @xmath75 is of type @xmath15 .
each clone @xmath77 is defined to be the same type as the model it is mapped to by @xmath41 .
clones at higher levels are defined to be of the same type as the model they are mapped to by compositions of the similarity maps , @xmath41 .
each @xmath41 is a similarity map from a level-1 clone to a model .
there is a lot of information we need to convey with the label @xmath41 , such as which model the clone @xmath77 is mapped to and which it is contained in . for the sake of clarity we use only one subscript throughout most of this paper . in the next two paragraphs
we include an explanation of the rules for compositions of these similarity maps . to do this we introduce three indices for each map which we shall subsequently omit in later sections .
we define @xmath85 where @xmath86 , @xmath87 and @xmath88 . here
@xmath89 is the type of model the level-1 clone @xmath77 is contained in , @xmath90 is the type of the level-1 clone @xmath77 , and @xmath91 is the number of level-1 clones of type @xmath90 , contained in model @xmath89 . models , level-1 clones and similarity maps .
( type 1 is dark , type 2 is light.),width=396,height=113 ] the compositions of these similarity maps are called _ clone - model maps_. they carry the restriction that if @xmath92 is composed with @xmath93 to form @xmath94 then @xmath89 must be equal to @xmath95 . for example
@xmath96 is not a possible composition , but @xmath97 is .
also , the index in the @xmath90 position of the first map and the index in the @xmath89 position on the last map must be the same .
for example , @xmath98 is not defined but @xmath99 is defined .
the domain of each clone - model map is by definition a _
clone_. a _ level - k _ clone is the domain of a composition of @xmath100 of the clone - model maps .
below are three pictorial representations of a single multi - model cantor set in @xmath62 .
this multi - model cantor set is given by the dynamical system @xmath101 .
the models of a multi - model cantor set ( shown fully in figure [ 2dmulti4]).,height=207 ] the models and level-1 clones of figure [ 2dmulti4].,height=207 ] a multi - model cantor set in @xmath62 with clone structure @xmath102.,height=207 ] figure [ 2dmulti4 ] is a multi - model cantor set in @xmath62 .
the models , seen in figure [ 2dmulti6 ] , are @xmath57 - a square , @xmath103 - a circle and @xmath104 - a rectangle .
the level-1 clones @xmath105 are contained in the models . these can be seen more clearly in figure [ 2dmulti5 ] .
we can see that the similarity maps from the level-1 clones onto the models are quite simple .
if we assume the top left vertex to be the origin then clone @xmath106 is mapped onto @xmath57 by a stretch factor of 2 .
clone @xmath107 maps onto @xmath103 by a stretch factor of 2 and a translation .
clone @xmath108 maps onto model @xmath104 by a stretch factor of 2 and a translation .
clone @xmath109 maps onto @xmath57 by a stretch factor of 4 and a translation .
clone @xmath110 maps onto @xmath104 by a stretch of factor 4 and a translation .
clone @xmath111 maps onto @xmath57 by a stretch of factor 2 and a translation .
clone @xmath112 maps onto @xmath103 by a rotation by @xmath70 and a translation and @xmath113 maps onto @xmath57 by a stretch of factor 2 and a translation . +
* definition . * given a multi - model cantor set @xmath8 with @xmath53 models , we define for each @xmath9 an @xmath114 matrix @xmath11 as follows .
the @xmath12 entry is the sum of the @xmath13-powers of the inverse scale factors of the clone maps of the level-1 clones of type @xmath14 , contained in model @xmath15 .
recall , @xmath115 is the number of type @xmath14 clones contained in model @xmath15 .
so each entry is of the form @xmath116 where @xmath117 is the inverse of the scale factor of the revelant clone map .
a multi - model cantor set and its corresponding matrix.,width=396,height=132 ] @xmath118 + figure [ matrix ] gives an example of a multi - model cantor set and its corresponding matrix .
the scale factors of the similarity maps are given on the diagram .
* definition . * given a multi - model cantor set @xmath0 with @xmath53 models and a finite collection of disjoint clones @xmath119 , where @xmath14 is the type of clone and @xmath120 where @xmath121 is the number of type @xmath14 clones in j. then given @xmath9 we define a vector @xmath122 by @xmath123 we say @xmath124 is the _ d - quantity _ of the collection of clones @xmath125 . +
* definition .
* given a finite collection of disjoint clones @xmath125 and @xmath126 let @xmath127 be the collection of clones obtained by subdividing , @xmath100 times , each clone in @xmath125 .
( by subdividing once we mean replacing each clone by the clones contained in it at relative level 1 . )
[ power ] given a multi - model cantor set @xmath8 and given @xmath9 let @xmath128 be the the multi - model cantor set matrix defined above . given a finite collection of clones @xmath125 then @xmath129 first observe that if @xmath130 and @xmath131 then @xmath132 thus it suffices to prove the result when @xmath125 is a single clone .
the case @xmath133 follows from the definition of @xmath11 ( refer to figure [ matrix ] ) .
the inductive step follows from the observation that @xmath134 and applying the case @xmath133 to the collection @xmath135 [ mtothek ] given a multi - model cantor set @xmath51 with multi - model cantor set matrix @xmath11 , then @xmath136 is the multi - model cantor set matrix for @xmath137 ( observe that the cantor sets @xmath51 and @xmath138 are identical . )
* definition . *
a square matrix @xmath139 is _ irreducible _ if all its entries are non - negative and there exists some @xmath140 such that all entries of @xmath141 are positive .
( note this is not the same definition of irreducible used in representation theory ; namely that there are no invariant subspaces ) . in this paper
we study multi - model cantor sets with irreducible matrices .
the geometric significance of irreducibility , by corollary [ mtothek ] , is that a clone of each type is contained at some level within each model . here
are two figures which illustrate this .
the first is irreducible , the second is not .
a multi - model cantor set with irreducible matrix .
( each clone is contained in each model.),width=396,height=113 ] a multi - model cantor set with non - irreducible matrix .
( type 3 clones are not contained in models @xmath57 or @xmath103.,width=396,height=113 ] figure [ irreducible ] shows a multi - model cantor set with an irreducible matrix .
there are no zeros in the corresponding matrix , @xmath136 , for any @xmath140 . in figure [ reducible ]
we see a multi - model cantor set which does not have an irreducible matrix .
the corresponding matrix @xmath136 has zero entries in the @xmath142 row of both the @xmath143 and @xmath144 columns , for every @xmath140 .
in this section we prove theorem [ finite measure ] .
[ frob ] an @xmath10 irreducible matrix @xmath139 always has a real positive eigenvalue @xmath145 that is a simple root of the characteristic equation and is larger than the moduli of all the other eigenvalues . to this ` maximal ' eigenvalue @xmath145
there corresponds an eigenvector with strictly positive coordinates .
* definition . *
the _ frobenius eigenvalue _ of an irreducible matrix is the eigenvalue of largest modulus . from the frobenius theorem
we see that it is positive , real and has multiplicity 1 .
a _ frobenius eigenvector _ is a corresponding positive , real eigenvector .
the first step in the proof of theorem [ finite measure ] is to show that the frobenius eigenvalue @xmath17 , of the multi - model cantor set matrix @xmath11 , is a strictly decreasing continuous function of @xmath13 . from this
we shall see that there is a unique @xmath9 such that @xmath18 , and that this is the hausdorff dimension of the cantor set . we state some easily proved results on matrices without proof . +
* definition . *
the _ @xmath146-norm _ on @xmath73 is @xmath147 , given by @xmath148 [ prop 1 ] let @xmath139 be an @xmath10 irreducible matrix with frobenius eigenvalue @xmath145
. then there exists @xmath149 such that for all @xmath100 and for all vectors @xmath19 , @xmath150 furthermore when @xmath151 there is a strictly positive matrix @xmath152 such that @xmath141 converges to @xmath152 as @xmath100 tends to @xmath153 .
* definition .
* we put a partial ordering on real matrices .
we say matrix @xmath65 is _ less than _ matrix @xmath59 if each entry in @xmath65 is less than the corresponding entry in @xmath59 .
so @xmath154 if @xmath155 for all @xmath14 and @xmath15 .
[ eigen ] if @xmath145 is an eigenvalue of @xmath139 with eigenvector @xmath19 then @xmath156 is an eigenvalue of @xmath141 with eigenvector @xmath19 .
note , if @xmath145 is the frobenius eigenvalue then it has largest modulus of all eigenvalues for @xmath139 so @xmath156 has largest modulus of all eigenvalues for @xmath141 .
therefore @xmath156 is the frobenius eigenvalue for @xmath141 .
[ bigger ] suppose @xmath157 , and @xmath158 is the frobenius eigenvalue for @xmath159 , then @xmath160 .
[ decreases ] suppose @xmath11 is an @xmath10 strictly positive matrix such that each entry is a strictly monotonically decreasing continuous function of @xmath13 .
let @xmath17 be the frobenius eigenvalue of @xmath11 for each @xmath13 . then as @xmath13 increases , @xmath17 decreases .
let @xmath161 .
each entry of @xmath11 is a strictly monotonically decreasing function of @xmath13 so @xmath162 .
then by lemma [ bigger ] we have @xmath163 .
suppose @xmath11 is an @xmath10 strictly positive matrix where each entry is a strictly monotonically decreasing continuous function of @xmath13 , and @xmath17 is the frobenius eigenvalue of @xmath11 for each @xmath13 .
then @xmath164 such that @xmath165 is continuous .
let @xmath166 be the characteristic polynomial of @xmath11 .
the coefficients of @xmath166 are polynomial functions of the entries of @xmath11 .
since polynomials are continuous and the entries of @xmath11 are continuous functions of @xmath13 , it follows that the coefficients of @xmath166 vary continuously with @xmath13 .
frobenius gives us @xmath17 which is a real , positive root of @xmath166 with multiplicity 1 .
without loss of generality we can assume @xmath167 and @xmath168 for some small @xmath169 .
now consider @xmath170 where @xmath171 for some small @xmath172 , then @xmath173 and @xmath174 by continuity of coefficients of the characteristic polynomial .
then by the intermediate value theorem @xmath175 for some @xmath176 .
we have shown that @xmath170 has a root @xmath177 in the interval @xmath178 .
we now complete the proof that @xmath179 is continuous .
we argue by contradiction .
if as @xmath180 we have @xmath181 then , since @xmath17 is monotonic in @xmath13 , there is an @xmath182 and a sequence @xmath183 such that @xmath184 with @xmath185 since @xmath186 is the largest root we know that @xmath187 thus @xmath188 .
for @xmath53 large we have @xmath189 so by corollary [ decreases ] we have @xmath190 .
therefore the @xmath186 are contained in a compact interval and so there exists a convergent subsequence .
re - label the convergent subsequence , then @xmath191 * claim . *
the limit @xmath192 is a root of @xmath166 .
_ the sequence @xmath193 and so @xmath194 because the coefficients of @xmath166 are continuous functions of @xmath13 .
the characteristic equation @xmath195 has a root @xmath186 for all n
so @xmath196 this proves the claim .
the claim contradicts @xmath17 being the largest root of @xmath166 .
the contradiction implies that @xmath177 is the largest root of @xmath170 and so @xmath197 .
therefore @xmath17 varies continuously with @xmath13 and hence @xmath165 is continuous . in the following lemmas
we shall suppose that the irreducible matrix @xmath11 is strictly positive .
this can be justified as follows .
given @xmath7 and an integer @xmath198 the multi - model cantor sets @xmath51 and @xmath138 are identical .
however their clone structures differ .
the corresponding multi - model cantor set matrices are respectively @xmath11 and @xmath136 by corollary [ mtothek ] .
the hypothesis that @xmath11 is irreducible ensures that we may choose @xmath100 sufficiently large so that @xmath136 is strictly positive . therefore replacing @xmath7 by @xmath199
justifies the supposition .
suppose @xmath11 is a multi - model cantor set matrix and @xmath17 is the frobenius eigenvalue for each @xmath13 .
then there is a unique @xmath9 such that @xmath18 .
recall that the inverse scale factors of clone maps satisfy @xmath200 . as @xmath13 approaches @xmath153
, we have that @xmath201 gets close to @xmath202 , for all @xmath12 .
this implies @xmath11 approaches the zero matrix and so @xmath17 tends to @xmath202 by lemma [ bigger ] . when @xmath203 , the sum @xmath204 for all @xmath12
. thus each of the entries of the matrix @xmath205 are at least @xmath206 .
let @xmath19 be the eigenvector for @xmath207 then @xmath208 note , @xmath19 is the frobenius eigenvector of @xmath205 and hence @xmath209 for all @xmath14 , therefore @xmath210 , for all @xmath15 .
this implies @xmath211 and hence @xmath212 .
the function @xmath165 is continuous , so by the intermediate value theorem @xmath213 for some value of @xmath9 .
the function @xmath179 is strictly monotonic by lemma [ bigger ] .
therefore this @xmath13 is unique .
[ b ] suppose @xmath139 is an irreducible non - negative @xmath10 matrix with frobenius eigenvalue @xmath206 .
then there is a @xmath149 such that for all @xmath214 and for all vectors @xmath19 in @xmath73 we have @xmath215 the following gives the hausdorff measure of every model in a multi - model cantor set .
a clone is a scaled down copy of a model and a clopen is a finite union of clones .
thus the hausdorff measure of every clopen can be determined from the following result .
[ finite measure ] suppose @xmath9 and @xmath0 is a multi - model cantor set with matrix @xmath11 and @xmath17 is the frobenius eigenvalue for each @xmath11 .
let @xmath13 be chosen such that @xmath18 .
then the hausdorff measure of @xmath0 is finite and non - zero in this dimension .
therefore @xmath13 is the hausdorff dimension of @xmath0 .
also , let @xmath19 be the frobenius eigenvector of @xmath11 such that @xmath20 then @xmath21 . by lemma [ a ] , if we show that @xmath216 then @xmath13 is the hausdorff dimension of @xmath0 . to show @xmath217 it suffices to show there is a @xmath24 such that for all @xmath218 there is an open cover @xmath219 of @xmath0 such that @xmath5 for all @xmath4 in @xmath219 and @xmath220 given @xmath3 there is a @xmath140 such that the diameter of every level - k clone is less than @xmath221 .
let @xmath219 be the set of level - k clones .
then @xmath219 is an open @xmath221-cover of @xmath0 . if @xmath128 and if @xmath222 are the level-0 models then by lemma [ power ] the @xmath13-quantity of @xmath219 is @xmath223 now @xmath141 converges to @xmath152 as @xmath224 by lemma [ prop 1 ] . hence @xmath225
if we take @xmath226 then we are done . to show @xmath227 it suffices to show there is an @xmath182 such that for every open cover @xmath219 of @xmath0 we have @xmath228 given an open set @xmath4 , let @xmath229 be the smallest clone containing @xmath4 . by lemma
[ 10.3 ] there is a constant @xmath230 depending only on the cantor set and @xmath7 such that @xmath231 hence @xmath232 therefore it suffices to show that there is an @xmath233 such that for every cover @xmath219 of @xmath0 by clones we have @xmath234 then @xmath235 satisfies the above .
given a cover of @xmath0 by clones , since @xmath0 is compact there is a finite subcover .
we may choose the subcover so that no clone in the subcover is contained in any other clone in the subcover .
if two clones intersect then one is contained in the other .
hence we may choose a subcover consisting of finitely many disjoint clones .
therefore it suffices to prove inequality ( [ eta ] ) for @xmath236 a finite cover of @xmath0 by disjoint clones .
the clones in @xmath219 may be at different levels .
let @xmath237 be the cover obtained by subdividing the clones in @xmath236 so that they are all at the same level , @xmath100 .
* there is a @xmath149 such that for all @xmath100 @xmath238 given the claim we see that the right hand side is the sum over all level - k clones .
we have seen above that as @xmath224 this sum converges to @xmath239 .
thus for @xmath100 large we have @xmath240 thus we have shown that @xmath241 satisfies ( [ eta ] ) .
it remains to prove the claim .
_ proof of claim . _
it suffices to show this when @xmath219 contains only one clone @xmath242 . by lemma [ b ]
, there is a @xmath149 such that @xmath243 for all positive vectors @xmath19 in @xmath73 .
let @xmath242 be a clone of type @xmath244 then @xmath245 is the @xmath13-quantity of the clone .
the clone @xmath242 subdivided @xmath100 times forms the collection of clones @xmath127 with @xmath13-quantity @xmath246 .
then by lemma [ b ] we have
@xmath247 this proves the claim .
this proves that the hausdorff measure of @xmath0 is finite and non - zero in this dimension . therefore , by lemma [ a ] , we see that @xmath13 is the hausdorff dimension .
let @xmath19 be the vector where @xmath248 is the hausdorff measure of the level-0 model @xmath40 .
the @xmath12 entry of the transposed matrix @xmath249 is the sum of the @xmath13-powers of the inverse scale factors of the clone - model maps from the type @xmath15 clones to the type @xmath14 models .
then the @xmath14 entry of @xmath250 is @xmath251 by lemma [ multiply ] , this is the sum of the hausdorff @xmath95-measures of all the level-1 clones contained in model @xmath14 .
hausdorff measure is @xmath252-additive so the hausdorff @xmath95-measure of model @xmath40 must equal the measure of the level-1 clones contained in @xmath40
. therefore @xmath253 this implies @xmath19 is an eigenvector with eigenvalue 1 .
in this section we shall generalize some results by cooper @xcite on clone cantor sets to the case of multi - model cantor sets .
the main result is theorem [ 10.6 ] which states that given a bilipschitz map between multi - model cantor sets then the map is measure linear on some clone . + * definition . * given a subset @xmath65 of a clone cantor set @xmath0 , the _ separation of @xmath65 _ is the minimum distance between @xmath65 and @xmath254 .
@xmath255 by convention , @xmath256 .
the _ relative separation of @xmath65 _ is the quotient of the separation of @xmath65 by the diameter of @xmath65 : @xmath257 as we discuss concepts related to cantor sets we will often want to consider clones after `` rescaling '' . applying a similarity map with stretch factor @xmath22 to clones increases the diameter of the clones by a factor @xmath22 and also increases the diameter of the gaps between clones by the same factor .
thus the quotient of these values remains constant .
relative separation is therefore an invariant factor under rescaling . + * definition . *
the _ minimum separation of c _ is denoted @xmath258 we now generalize some of the results from a paper written by cooper @xcite on clone cantor sets .
we have made the necessary changes and additions to make these work in the case of the multi - model cantor set .
we have rewritten the statements of the lemmas and the main theorem from this paper and have summarized each .
we have outlined the main points of the proofs .
the reader should refer to @xcite for further details .
lemma [ 10.2 ] includes a change of wording to signify that @xmath7 is a map from a clone to the level-0 model of the same type .
here we use the definition of clone - model map given in section [ multi - model ] .
lemmas [ 10.3 ] , [ 10.4 ] and [ 10.5 ] now involve an extra case where a clopen is not contained in any model and theorem [ 10.6 ] follows from these .
this lemma tells us that the separation of a clone approximates its diameter .
the replacement of the clone map by a clone - model map suffices to make the proof from @xcite apply to the multi - model case .
we now sketch this .
there are only a finite number of clones contained in the cantor set of more than a certain diameter .
two points are chosen , of minimal distance apart , one inside clone @xmath65 and the other in @xmath254 .
rescaling by a clone - model map does not change relative separation , so we can assume the two points are contained in different level-1 clones and that the distance between the two points is no less than the minimum separation of @xmath0 .
this now gives a lower bound on the diameter of @xmath60 , the smallest clone which properly contains @xmath65 .
there are only a finite number of possibilities for @xmath60 due to the lower bound on its diameter .
therefore there are a finite number of options for @xmath65 , as @xmath65 is at relative level 1 to @xmath60 .
since there are only a finite number of choices for @xmath65 then there are only a finite number of possibilities for @xmath262 .
@xmath263 [ 10.3 ] given a multi - model cantor set c there is a constant @xmath90 with the following property .
let b be any subset of c of positive diameter and let d be the smallest clone of c containing b ( or choose d = c if there is no such clone ) . then @xmath264 given a multi - model cantor set @xmath0 , we note that not every subset of @xmath0 is contained in a clone . however , if @xmath59 is not contained in any clone then @xmath265 . taking @xmath60 to be the whole cantor set
then we already have @xmath266 and the result follows as below .
if @xmath59 is contained in a clone then rescale @xmath60 onto the appropriate level-0 model @xmath75 , using a clone - model map @xmath7 .
after rescaling , @xmath59 must contain points in at least two level-1 clones by choice of @xmath60 .
therefore the diameter of @xmath59 must be at least as great as the minimum separation of @xmath0 . now rescaling multiplies the diameter of all clones by the same constant
so the ratio of the diameter of @xmath60 to @xmath59 is the same as the ratio of the diameters after rescaling .
this is then no greater than the ratio of the diameter of @xmath75 to the minimum separation of @xmath0 .
therefore @xmath267 so if we let @xmath268 we get the desired result .
@xmath263 [ 10.4 ] given @xmath269 , @xmath169 and multi - model cantor sets
c , @xmath270 there is a constant m with the following property .
let f be any k - bilipschitz map of c onto a clopen in @xmath270 , and suppose that @xmath271 .
let a be any clone of c. if @xmath272 is the smallest clone of @xmath270 containing f(a ) and @xmath273 is the clone map taking @xmath272 onto @xmath270 then there are at most @xmath274 possibilities for the image @xmath275 as a subset of @xmath270 .
this is also true if @xmath276 is not contained in any clone of @xmath270 with @xmath273 taken to be the identity map .
this lemma says that given a @xmath22-bilipschitz map from @xmath0 to @xmath270 and a clone @xmath65 in @xmath0 , up to rescaling in @xmath270 , there are only a finite number of possibilities for the image of @xmath65 .
the proof given in @xcite does not consider the case where the clopen @xmath276 is not contained in any clone .
suppose the clopen @xmath276 is not contained in any clone .
then there exist points @xmath2 and @xmath277 in @xmath276 such that @xmath2 and @xmath277 are in different models , i.e. level-0 clones .
this implies @xmath278 .
the rest of the proof in @xcite then applies but with @xmath273 replaced by the identity map .
we now sketch this proof .
the relative separation is not affected by rescaling . given a clone @xmath65 in @xmath0 , then the diameter of @xmath276 is bounded above by a factor of @xmath22 times the diameter of @xmath65 because @xmath279 is @xmath22-bilipschitz .
the separation of @xmath276 is bounded below by a factor of @xmath280 times the separation of @xmath65 for the same reason .
hence the relative separation of @xmath276 is bounded below by some positive constant .
we then rescale @xmath276 using a clone - model map @xmath273 and find that the diameter of @xmath281 is bounded below by the minimum separation of models in @xmath270 and as relative separation is unchanged by rescaling we find the separation of @xmath281 is also bounded below .
we then choose @xmath59 to be the biggest clone in @xmath281 and let @xmath60 be the smallest clone which properly contains @xmath59 ( or all of @xmath270 ) .
the diameter of @xmath60 is bounded below by the separation of @xmath281 and so there are only finitely many posibilities for @xmath60 .
clone @xmath59 is one level higher than clone @xmath60 and hence there are only finitely many possibilities for @xmath59 .
the number of maximal clones in @xmath281 is bounded above because the diameter of each is bounded below .
since there are a finite number of possibilities for the image of each of these clones we have a finite number of possibilities for @xmath281 .
@xmath263 [ 10.5 ] given @xmath282 , @xmath169 and multi - model cantor sets c , @xmath270 there is a finite set s of positive real numbers with the following property .
suppose that f is any k - bilipschitz map of c onto a clopen in @xmath270 , and suppose that @xmath271 .
if b is any clone of level n in @xmath0 containing a clone d of level n+1 then @xmath283 this may be interpreted as saying that non - linearity , at the level of measure theory , comes in discrete `` packets '' .
a bilipschitz map need not distort the measure of all clones by the same multiplicative factor but , if @xmath60 is a clone contained in @xmath59 of relative level 1 , then the ratio of the masses of the images of @xmath59 and @xmath60 can be one of only finitely many possible numbers . in generalizing the proof to the case of multi - model cantor sets we need to consider the possibility that @xmath284 is not contained in any clone .
by lemma [ 10.4 ] we already know there are a finite number of possibilities for @xmath284 . we need to show there are only finitely many possibilities for @xmath285 and the result will follow . * case 1 . *
suppose @xmath285 is contained in some clone @xmath286 .
we know that @xmath287 where @xmath75 is a model and @xmath77 is a level-1 clone .
hence this quotient is equal to one of a finite number of possible positive scale factors and so is bounded away from 0 .
now @xmath284 is not contained in a model so @xmath288 and hence @xmath289 as @xmath279 is @xmath22-bilipschitz .
we have that @xmath290 approximates @xmath291 by lemma [ 10.3 ] and so approximates @xmath292 because @xmath279 is @xmath22-bilipschitz .
hence @xmath293 and so is bounded away from 0 and thus @xmath294 is bounded away from 0 .
there are only a finite number of clones this large so we have only finitely many choices for @xmath286 .
we map @xmath286 onto the appropriate level-0 model @xmath75 by a clone - model map @xmath295 , ( using @xmath296 and @xmath297 identity if level @xmath298 )
. then @xmath299 is one of only finitely many possibilities by lemma [ 10.4 ] and since there are only a finite number of choices for @xmath295 then there are only finitely many possibilities for the clopen @xmath285 .
if @xmath284 is contained in a clone then the rest of the proof for lemma [ 10.5 ] follows from the observation that there are a finite number of possibilities for the rescaled image of clone @xmath59 , and the rescaled image of clone @xmath60 , where @xmath60 contains @xmath59 at relative level 1 .
therefore there are only a finite number of values for @xmath300 we take @xmath273 to be the identity map if @xmath284 is not contained in any clone . in either case
the map @xmath273 is a similarity map with stretch factor @xmath301 we have , using lemma [ multiply ] , @xmath302 thus there are only a finite number of possibilities for this ratio . @xmath263 + * definition . *
given a @xmath22-bilipschitz map @xmath303 , the _ mass ratio _ ( @xmath304 ) of a clone @xmath65 in @xmath0 is defined to be @xmath305 here we see that when a clone @xmath60 and a clone @xmath310 contained within it at relative level 1 are mapped into a multi - model cantor set with a @xmath22-bilipschitz map then the quotient of the mass ratios of these clones is restricted to some finite set .
+ * definition . *
a map @xmath311 between metric spaces is _ measure linear _ , i.e. has constant radon - nikodym derivative , if for some constant @xmath22 , @xmath312 for all measurable subsets @xmath65 of @xmath26 .
this is a weak generalization of linearity .
linearity refers to metric whereas this concept refers to measure . for lebesgue measure on the real line ,
linear and measure linear are the same for continuous functions .
[ 10.6 ] suppose that c , @xmath270 are multi - model cantor sets and that f is a @xmath22-bilipschitz map of c onto a clopen subset of @xmath270 .
then @xmath0 and @xmath270 have the same hausdorff dimension .
furthermore there is a clopen a in c such that the restriction @xmath313 of f to a is measure linear with respect to hausdorff measure .
we sketch the proof from @xcite .
the map @xmath279 is @xmath22-bilipschitz so the mass ratio function is bounded above .
thus there exists a supremum for the possible mass ratios .
let @xmath65 be a clone contained in @xmath0 with mass ratio within @xmath221 of this supremum .
first note that if the mass ratios of all clones contained in @xmath65 were equal , then we would already have a measure linear map on @xmath65 .
also if each clone contained in @xmath65 had mass ratio smaller than that of @xmath65 then the union of these clones would also have smaller mass ratio which is a contradiction .
so assume there is a clone @xmath310 of minimum level , with mass ratio greater than that of @xmath65 .
we call this a `` greedy '' clone .
we then , by corollary [ massratio ] , have only a finite number of possibilities for the quotient of mass ratios of @xmath310 and the smallest clone @xmath60 which contains it .
if the mass ratio of @xmath60 were larger than the mass ratio of @xmath65 we would have chosen this for @xmath310 as it has lower level .
on the other hand the mass ratio of @xmath60 can not be smaller than the mass ratio of @xmath65 else there exists a greedy clone at the same level as @xmath60 .
therefore the mass ratio of @xmath60 must be equal to the mass ratio of @xmath65 .
this implies that the quotient of mass ratios of @xmath310 and @xmath60 must be greater than 1 .
thus @xmath314 is in a finite set of numbers bigger than 1 .
this implies that the mass ratio of @xmath310 is larger than the supremum of mass ratios , which is of course a contradiction and so proves that there could not have been a `` greedy '' clone .
therefore the map must have been measure linear on @xmath65 .
@xmath263 in @xcite , cooper defines various invariants of clone cantor sets under bilipschitz maps .
we say two sets of real numbers @xmath65 and @xmath59 are _ similar _ if there exist non - zero scalars @xmath315 such that @xmath316 let @xmath0 be a multi - model cantor set with hausdorff measure @xmath317 where @xmath95 is the hausdorff dimension .
we define the _ clopen invariant _ of an arbitrary subset @xmath59 of @xmath0 as the similarity class of the countable set of real numbers @xmath318 .
the clone - model map from a to the model @xmath75 shows that @xmath65 has the same clopen invariant as model @xmath75 .
similarly @xmath59 has the same clopen invariant as @xmath40 for some @xmath14 .
a scaled down copy of @xmath75 is contained in @xmath40 for all i hence the clopen invariant for the model @xmath75 is similar to that for every other model .
we can now define the clopen invariant of a multi - model cantor set .
let @xmath0 be a multi - model cantor set with hausdorff measure @xmath317 where @xmath95 is the hausdorff dimension .
we define the _ clopen invariant _ of @xmath0 as the similarity class of the countable set of real numbers @xmath319 for some @xmath15 .
the choice of @xmath15 is irrelevant by lemma [ c ] . by theorem [
10.6 ] , there is a clopen , hence a clone @xmath65 in @xmath0 with a measure linear map of @xmath65 onto a clopen in @xmath270 .
the clone - model map from @xmath65 to @xmath75 for some @xmath15 shows that @xmath65 has the same clopen invariant as @xmath75 and hence @xmath0 by lemma [ c ] .
let @xmath89 be the radon - nikodym derivative of the measure linear map of @xmath65 into @xmath270 .
the image of a clopen @xmath59 in @xmath65 is a clopen in @xmath270 with @xmath89 times the measure of @xmath59 . by the same arguement as above
, this clopen invariant is the same as the clopen invariant for @xmath270 . using that the inverse of a bilipschitz map is bilipschitz
, we obtain the inverse relationship .
primary subject : 37d40 dynamical systems of geometric origin .
+ secondary subject : 28a80 fractals + e. cockerill + department of mathematics + university of california + santa barbara , ca 93106 + [email protected] | in this paper we define a new class of metric spaces , called multi - model cantor sets .
we compute the hausdorff dimension and show that the hausdorff measure of a multi - model cantor set is finite and non - zero .
we then show that a bilipschitz map from one multi - model cantor set to another has constant radon - nikodym derivative on some clopen .
we use this to obtain an invariant up to bilipschitz homeomorphism . |
the discovery by hst and large ground - based telescopes , of a large population of superstar clusters ( sscs ) in a large variety of galaxies ( see ho 1997 , martn hernndez et al . 2005 , and melo et al . 2005 ) , has driven the community to speculate about the impact that these new units of star formation may have on the ism of their host galaxies .
sscs with a mass , m@xmath1 , ranging from a few @xmath2 m@xmath3 to up to 6 @xmath4 m@xmath3 ( see walcher et al .
2004 ; pasquali et al .
2004 ) all within a small volume of radius 3 to 10 pc , are indeed some of the most energetic entities within galaxies .
their large uv photon output and mechanical energy are now believed to be the largest negative feedback agents in starburst galaxies , leading not only to a large - scale structuring of the ism and to limit star formation , but also to be the agents capable of establishing as in m82 a supergalactic wind , thereby removing processed material from galaxies and causing the contamination of the igm ( tenorio - tagle et al .
2003 ) . in the present series of papers however
, we have shown that there is a threshold limit that massive and energetic compact clusters may cross to find out that radiative cooling inhibits their stationary outflow condition ( silich et al . , 2004 ) and then the matter ejected by the stellar sources , unable to escape , is to accumulate within the star cluster volume ( tenorio - tagle et al . 2005 ) .
the facts in such cases are that radiative cooling drastically diminishes the sound speed and the pressure gradient within the thermalized ejected matter , inhibiting the possibility of a wind .
radiative cooling upsets the stationary condition in which the deposited matter ( @xmath5 ) has to equal , at all times , the amount of matter streaming out of the sc volume , @xmath6 , where @xmath7 is the amount of energy injected by stellar winds and sne , @xmath8 is the local sound speed at the sc surface and @xmath9 is the resultant wind terminal speed in the absence of radiative cooling .
as soon as this happens , the ejecta begins to accumulate , promoting larger densities and an even faster cooling within the ssc volume .
the location of the threshold line in the @xmath7 _ vs _ size ( @xmath10 ) diagram ( see figure 1 ) , the line that defines whether or not a wind is inhibited , depends on several variables .
it depends on the size of the star - forming region ( @xmath10 ) , the metallicity of the ejected gas , which has a strong impact on the cooling curve ( see paper iii ) .
it also depends on the assumed @xmath11 or adiabatic terminal speed ( @xmath12 ) of the wind .
the latter is also bound to the usual assumption that the energy deposited by sn is always @xmath13 erg but the mass of potential sne within the cluster ranges from , say , 100 m@xmath3 to 8 m@xmath3 and so the injection speed ( similar to @xmath12 ) and the deposited amount of matter are also functions of time .
another factor that strongly affects the location of the threshold line is the thermalization efficiency ( @xmath14 ) which simply defines the fraction of the mechanical energy that can be evenly spread within the cluster volume .
estimates of @xmath14 by several authors lead to values between 1 ( chevalier & clegg ; 1985 ) to 0.03 ( see melioli & del pino 2004 and references therein ) and depends simply on the proximity of the sources undergoing winds and sne , which through radiation may reduce the amount of energy available after thermalization .
we have shown that there are three different types of solutions : sscs far away from the threshold line ( low mass , low energy clusters ) undergo a quasi - adiabatic evolution well described by the chevalier & clegg ( 1985 ) and cant et al .
, ( 2000 ) solution .
more energetic clusters are to have strongly radiative winds .
cooling hardly affects their velocity ( @xmath15 ) and density distribution ( @xmath16 ) , but their temperature instead of falling as @xmath17 , it falls rapidly to @xmath18 k close to the sc boundary and the more so , the closer they are to the threshold line .
the strongly radiative winds around such clusters lead , compared to the adiabatic solution , to very much reduced x - ray envelope sizes .
the third solution is for clusters above the threshold line .
these would have their winds inhibited and as shown below , this turns them into very efficient positive feedback star - forming agents . here
we follow the evolution of a massive and compact cluster and show the fate of the mass reinserted back , through winds and sne , into the sc volume .
section 2 and 3 give a full description of ssc as positive feedback agents and section 4 summarizes our conclusions .
here we follow the evolution of a stellar cluster , assumed to be the result of an instantaneous or coeval burst of star formation .
the cluster has an initial total mass in stars ( @xmath19 ) equal to 10@xmath20 m@xmath3 within an @xmath10 = 5 pc . we assume a standard salpeter imf and an upper and lower mass limits equal to 100 m@xmath3 and 1 m@xmath3 , respectively .
the mechanical energy of the cluster @xmath21erg s@xmath22 places our example cluster in the zone where stationary superwinds are inhibited ( see figure 1 ) and thus the mass returned by the stars ( @xmath5 ) begins immediately to accumulate within the star cluster volume , favoring an even faster cooling .
figure 2 shows how the density of the accumulating gas ( @xmath23 ; where @xmath24 the evolution time ) , grows as a function of time within the ssc volume .
following this trend , clearly there is a moment when the density of the accumulating gas will exceed the gravitational instability criteria @xmath25 where @xmath26 is the jeans density and @xmath27 and @xmath28 are the boltzmann constant , gravitational constant , mean mass per particle and the ratio of specific heats ( 5/3 ) , respectively and @xmath29 is the ssc radius in pc units . at that moment , when @xmath30 collapse will inevitably proceed .
the initial ample supply of uv photons ( @xmath31 = 10@xmath32 photons s@xmath22 , see figure 3a ) , exceeds at first the number of recombinations within the volume occupied by the reinserted gas and the resultant hii region , given the large metallicities of the ejecta , is here assumed to rapidly approach an equilibrium temperature @xmath33 a few @xmath34 k. at these temperatures the sound speed ( @xmath35 10 km s@xmath22 ) remains well below the escape speed ( @xmath36 km s@xmath22 ) and the reinserted gas would inevitably continue to accumulate to rapidly ( within @xmath37 yr ) reach the value of the jeans density for a gas at say , @xmath38 = 5000 k , and collapse into a new stellar generation .
the event gives rise to a new phase of matter accumulation , which once more will rapidly approach @xmath26 ( for @xmath39= 5000 k ) and undergo collapse within a free - fall time , of the order of 10@xmath40 yr , while transforming @xmath41 m@xmath3 into stars .
all stellar generations resultant from mass accumulation within the ssc volume have here been assumed to also acquire a salpeter imf with similar upper and lower mass limits as those imposed to the main superstar cluster , and their resultant properties ( mechanical energy and uv photon output ) have been added to those produced by the main cluster . a few , almost identical , stellar generations are expected from the accumulation process ( solid rising lines in figure 2 ) , every time that the accumulated gas density @xmath42 reaches @xmath43 ( dashed line in figure 2 ) .
the situation changes slightly when the number of ionizing photons ( @xmath31 ) , despite the added contribution of secondary stellar generations , becomes insufficient to fully ionize the accumulated matter within the star cluster volume .
this is due to the evolution of the main cluster , whose uv photon output begins to fall as @xmath44 after @xmath45 3.5 myr ( dotted line in figure 3a ) .
figure 2 shows @xmath46 ( thin solid line ) , the maximum density within the accumulating volume that can be supported fully ionized by the uv radiation produced by the evolving cluster ( @xmath47 ; where @xmath48 and @xmath49 , the recombination coefficient to all levels above the ground level = 2.59 @xmath50 @xmath51 s@xmath22 ) . during the accumulation process ,
once @xmath42 exceeds @xmath46 , the ionized volume begins to shrink to end up as a collection of ultra compact hii regions around the most massive stars left within the cluster , while the bulk of the ejected material , now recombined , continues to cool , approaching rapidly ( within a time - scale of less than 1000 yr ) a temperature @xmath45 100 k. matter is at all times uniformly replenished within the whole ssc volume , and thus the gas density presents an almost uniform value .
however , the accumulating gas now has two different temperatures ( @xmath38 and 100 k ) and as @xmath42 grows and the fraction of the ionized volume ( @xmath52 ) shrinks , the size of cold condensations ( at 100 k ) able to become gravitationally unstable and their free - fall time also become smaller .
the drop in the number of ionizing photons ( figure 3a ) and the consequent growth of the neutral volume ( figure 3b ) lead then to a second important condition in which the characteristic accumulation time @xmath53 becomes equal to the free - fall time @xmath54 .
this condition defines @xmath55 , the density above the jeans instability limit for a neutral condensation at 100 k ( @xmath26(100k ) ) : @xmath56^{1/3 } r_{sc}^{-2 } = \frac{1.15 \times 10^{-18}}{(1 - f_{hii})^{2/3 } } \left(\frac{\dot m}{1m_\odot yr^{-1 } } \right)^{2/3 } \left(\frac{r_{sc}}{1pc}\right)^{-2 } , \ ] ] and thus once @xmath42 ( see figure 2 ) becomes equal to @xmath55 , a new stationary solution becomes possible .
everything happens very rapidly , compared to the evolution time - scale of the parental cluster ( @xmath45 40 myr ) , and almost at the same time .
the ejected matter is thermalized within the ssc volume and immediately begins to cool . at the same time that it accumulates making cooling even faster .
this allows it to rapidly reach the required @xmath55 value , above the jeans instability limit , that warrants its collapse in a similar time - scale , while the collapsing material is replenished by the newly ejected matter .
when this happens , the mass deposition rate from the cluster becomes equal to the rate of star formation .
gravitational collapse and star formation within the star cluster volume and with the matter injected by all sources , drive in this way a new stationary condition through a new era of quasi - continuous star formation in which @xmath0 is equal now to the star formation rate ( sfr ) .
all new generations of stars are due to deposit their ejecta from winds and sne during @xmath57 myr , the expected duration of the type ii sn phase in coeval star clusters ( leitherer & heckman 1995 ) , and at a rate that depends only on the mass of the stellar generation .
re - processing leads to an escalating @xmath0 , as this is directly proportional to the total mass in stars @xmath58 created within the star cluster volume ; where @xmath59 marks the start of the quasi - continuous star - forming phase and @xmath19 accounts for the initial mass in stars and for all generations prior to the quasi - continuous star formation phase ( see figure 2 ) .
this promotes a larger star formation rate , and a faster fulfillment of the collapse conditions ( @xmath60 , and @xmath61 ) . under these conditions
the sfr is : @xmath62 this equation has an exponential solution : @xmath63 } , \ ] ] where the characteristic time @xmath64 .
equation ( [ sf.b ] ) remains valid during the supernovae phase of the parental cluster . within this time interval , the continuous re - processing of the ejected material drives , as shown in figure 4 , the star formation rate to grow exponentially : @xmath65 } .\ ] ] note that the characteristic time @xmath66 is defined by the coeval star cluster model and is applicable to both the initial cluster and also to every new generation of stars formed from the re - ejected material .
if one assumes that the energy and mass deposition rates remain constant during a star cluster life time and that @xmath12 , the resultant terminal speed in the absence of radiative cooling , equals 1500 km s@xmath22 for all clusters , upon normalizing to the energy deposition rate adopted for the parental 10@xmath20 m@xmath3 cluster ( @xmath67 erg s@xmath22 ) , then @xmath68 yr , where @xmath69 is the mass in each stellar generation , including the initial or parent cluster .
the total mass in stars born within the star cluster volume , over the characteristic time @xmath70 myr , is given by equation ( [ sf.b ] ) .
we have here shown that massive and compact coeval clusters instead of driving a superwind able to disperse the surrounding ism and even channel its way into the igm , events that have make them been regarded as negative feedback agents , they are in fact extreme examples of positive star formation feedback .
massive and compact coeval clusters appear in the @xmath71 _ vs _ size diagram above the threshold line , in the region where radiative cooling inhibits the development of stationary superwinds .
we have here shown that in such cases the matter reinserted , through stellar winds and supernovae , is unable to escape and that after a short phase of matter accumulation , a new stationary solution in which @xmath0 becomes equal to the sfr is rapidly met . a positive feedback condition in which new stellar generations result in situ , from the collapse of the matter reinserted by the sources evolving within the star cluster volume .
the secondary star formation process while causing a faster mass deposition rate , drives the sfr to grow from 0.1 to 0.25 m@xmath3 yr@xmath22 over the parent cluster supernova phase ( @xmath72 40 myr ) .
the continuous re - processing of the ejected material leads effectively to a continuous transformation of the high mass stars into a low mass ( @xmath73 8 m@xmath3 ) population , keeping constant the total mass of the stellar component . in this way
massive clusters may survive as gravitationally bound systems ( see discussion in dale et al .
2005 and martn - hernndez et al .
2005 and references therein ) .
a central issue , regarding ism studies , is the fact that the more massive and compact clusters ( as those detected by hst and large ground - based telescopes ) , are unable to generate superwinds and shed their metals into the ism or the igm .
their evolution leads to many stellar generations and thus to a mixture of stellar populations , all contaminated by the products from former stellar generations .
exacerbated episodes of star formation that leave no trace of their evolution in the ism .
clearly , if the most massive star formation events do not contribute to the chemical evolution of their parent galaxy , then several issues , such as the mass - metallicity relationship between galaxies , ought to be revised .
we thank our anonymous referee for comments and suggestions .
this study has been supported by conacyt - mxico , research grant 47534-f and aya2004 - 08260-co3-o1 from the spanish consejo superior de investigaciones cientficas .
cant , j. , raga , a.c . &
rodriguez , l.f .
2000 , apj , 536 , 896 chevalier , r.a . & clegg , a.w .
1985 , nature , 317 , 44 dale , j.e . ,
bonnell , i.a . ,
clarke , c.j .
& bate , m.r .
2005 , mnras , 358 , 291 ho , l. c. 1997 , rev.mexaa , conf .
ser . 6 , 5 leitherer , c. & heckman , t.m . , 1995 ,
apjs , 96 , 9 martn - hernndez , n.l . ,
schaerer , d. & sauvage , m. 2005 , a&a 429 , 449 melioli , c. & de gouveia del pino e. m. 2004 a&a 424 , 817 melo , v. p. , muoz - tun , c. , maz - apellniz , j. & tenorio - tagle , g. 2005 , apj , 619 , 270 pasquali , a. , gallagher , j. s. & de grijs , r. 2004 a&a 415 , 103 silich , s. , tenorio - tagle g. & rodrguez gonzlez , a. 2004 , apj , 610 , 226 ( paper ii ) tenorio - tagle , g. , silich , s. & muoz - tun , c. 2003 , apj , 597 , 279 tenorio - tagle , g. , silich , s. , rodrguez - gonzlez a. & muoz - tuon , c. , 2005 , apj 620 , 217 , ( paper iii ) walcher , c.j . , van der marel , r.p . , mclaughlin , d. , rix , h .- w . ,
bker , t. , hring , n. , ho , l.c . ,
sarzi , m. & shields , j.c .
2005 , apj , 618 , 237 | the force of gravity acting within the volume occupied by young , compact and massive superstar clusters , is here shown to drive in situ all the matter deposited by winds and supernovae into several generations of star formation .
these events are promoted by radiative cooling which drains the thermal energy of the ejected gas causing its accumulation to then rapidly exceed the gravitational instability criterion . a detailed account of the integrated ionizing radiation and mechanical luminosity as a function of time is here shown to lead to a new stationary solution . in this ,
the mass deposition rate @xmath0 , instead of causing a wind as in the adiabatic solution , turns into a positive feedback star - forming mode equal to the star formation rate .
some of the implications of this extreme positive feedback mode are discussed . |
in the limit of a large number of colours @xmath0 of qcd @xcite the baryon appears as a soliton @xcite . by now , various effective models of qcd in the low energy region have employed the soliton picture of the baryon : the skyrme model @xcite , the nambu - jona - lasinio ( njl ) model @xcite and the chiral quark soliton model ( cqsm ) @xcite .
the static properties ( mass , magnetic moment ) @xcite and the quark distribution functions @xcite of the baryon have been studied by means of these models . here , the baryon appears as a rotational band of the soliton @xcite .
the soliton takes the hedgehog shape and adiabatically rotates as a rigid body in flavor and real space .
although there is also an approach using the harmonic approximation of the meson fluctuations off the shape @xcite , in any case the stability of the shape is assumed . from the point of view of the constituent quark
, the assumption is justified in the case of the nucleon ( @xmath1 ) , since the hedgehog shape of the chiral meson fields is caused by the s state of the u , d quarks which have equal mass @xcite .
however it may be not so for the strange baryons ( @xmath2 , @xmath3 , @xmath4 , etc . ) consisting of quarks with different masses .
the inertial force in the body fixed frame of the soliton depends on the mass .
therefore , the hedgehog shape would be unstable .
the baryon as a nonrigid rotator has been studied with the skyrmion @xcite . there
, the effects of the coriolis force were neglected in the shape of the static soliton and included perturbatively in the state vector for the collective rotation . for the excited states of the baryon ,
there are several studies of this problem in the nonrelativistic quark model @xcite . in this article
we shall argue this problem for the octet baryon by means of the cqsm .
this model provides the simplest foothold to study the above problem from the point of view of the relativistic quark model .
the shape of the soliton can be determined self - consistently as the mean field potential for the quark .
thus the model can eliminate the influence of the artificial assumption that the mean field potential takes a spherical shape . in sec .
[ sec : model ] , after a brief review of the su(3 ) chiral quark soliton model we show the cranking method for the deformed soliton and the mean field approximation for the rotated system . section [ sec : perturb ] deals with the perturbative expansion of the effective action . at the the lowest order expansion
, we show the profile functions and the energy of the deformed classical soliton . for the higher order expansion
, we construct the hamiltonian due to the collective motion of the soliton and the canonical quntization of the collective variables . in sec .
[ sec : diagonal ] , first we deal with the kaon in the background soliton and the stability of the combined system of the soliton and the kaon .
next we give an outline of the diagonalization of the collective hamiltonian and show the results for the octet baryon . finally , in sec .
[ sec : discuss ] we summarize the results and discuss the relation to other papers .
the chiral quark model in case of the flavor su(3 ) is given by the following path integral @xcite .
@xmath5\ ] ] with @xmath6 where @xmath7 is the quark field and @xmath8 is the chiral meson field @xmath9 su(3 ) .
furthermore , @xmath10 is the dynamical quark mass and @xmath11 is the current quark mass matrix given by @xmath12 where @xmath13 are the gell - mann matrices , @xmath14 , and @xmath15 because @xmath16 . in this article
, we set @xmath17 mev , @xmath18 mev , and @xmath19 mev .
using the path integral formula for the quark field , we find @xmath20},\\ i s_{f}\left [ u \right ] & = & n_{c } \log\det \left(i\partial_{t}-h\right ) , \label{eq : logdet}\end{aligned}\ ] ] where @xmath0 is the number of colors and @xmath21 is the quark hamiltonian . for the chiral field @xmath22 , we postulate the so called cranking form @xcite : @xmath23 where @xmath24 is the static chiral meson field , @xmath25 describes the adiabatic rotation of the system in su(3 ) flavor space , and @xmath26 describes the spatial rotation .
we write @xmath24 as @xmath27 where @xmath28 is the radial component of the profile function , @xmath29 is a unit vector in the isospin space , and @xmath30 represents pauli matrices . for the flavor rotation we write @xcite @xmath31 where @xmath32 is the flavor su(2 ) rotation operator and @xmath33 rotates @xmath34 into the strange directions .
in particular , we parametrize @xmath35 as @xmath36 where @xmath37 and @xmath38 is the isodoublet spinor .
although eq .
( [ eq : a_param ] ) was motivated by the bound - state approach @xcite , we will not treat the locality of the kaon wave function in this article . from eqs .
( [ eq : logdet ] ) and ( [ eq : aua ] ) , we obtain @xmath39 where @xmath40 is the rotated hamiltonian and @xmath41 is given by @xmath42 here , the term proportional to @xmath43 breaks the su(3 ) symmetry and @xmath44 is the wigner _
function of @xmath33 in the adjoint representation .
we show some important components , @xmath45 where @xmath46 we now wish to discuss this model in the mean field approximation .
suppose that the collective motion @xmath47 and @xmath48 is quantized and @xmath49 as an eigenstate of the collective hamiltonian .
the operators @xmath50 and @xmath51 convert a non - strange quark into a strange one .
therefore , if points to a specific direction in the isospin space , @xmath52 have nonzero values . for a fixed @xmath53
, we have the following constraint for @xmath54 , @xmath55 intuitively , @xmath53 and @xmath54 represent the quantity of the strangeness , respectively , and the asymmetry of the strangeness coupled to the isodoublet in @xmath49 .
the expectation value @xmath56 may be approximated by @xmath57 with @xmath58 equation ( [ eq : mq_eff ] ) indicates that the u , d , s quarks mix with each other by the rotation in the strange direction and their masses are renormalized in the body fixed frame .
figure [ fig : meff ] shows the @xmath53 and @xmath54 dependencies of the effective quark masses @xmath59 . and @xmath54 dependencies of the effective quark mass . for @xmath53 dependence ,
we set @xmath60 , at which @xmath61 and @xmath62 take the same value . for @xmath54 , we set @xmath63.,title="fig : " ] +
we separate @xmath64 into an unperturbed part @xmath65 and a perturbing part @xmath66 as follows : @xmath67 we expand @xmath68 in powers of @xmath69 , @xmath70 , and @xmath66 around the eigenstate of @xmath71 @xcite . since @xmath65 contains the @xmath72 , @xmath65 breaks not only the su(3 ) symmetry but also su(2 ) .
thus it has only rotation symmetry around the third axis in isospin space .
when the isospin space is mapped into the real space , the configuration of the system has an axial symmetry with respect to the specific axis in real space .
the grand spin operator in the quark space is given by @xmath73 where @xmath74 , @xmath75 , and @xmath76 are the total , orbital , and spin angular momenta , respectively , and @xmath77 denotes the isospin .
the above argument is expressed as @xmath78 $ } } = 0 , \label{eq : k3_symmetry}\ ] ] where @xmath79 is the third component of @xmath80 and we choose the @xmath81 axis as the specific one .
we call this equation `` @xmath82 symmetry ''
. the isosinglet part of eq .
( [ eq : k3_symmetry ] ) means that the radial component of the profile function @xmath28 depends only on @xmath83 and @xmath84 of the polar coordinates . on the other hand ,
the isotriplet part means that the unit vector @xmath29 has axial symmetry about @xmath81 axis .
thus , the deviation from the hedgehog shape takes place only in the @xmath83 and @xmath84 directions . in actual calculation
, we introduce the tensor operators @xmath85 defined by the following equations , where @xmath86 and @xmath87 @xcite .
@xmath88 $ } } = k_{3 } z^{jlsi}_{kk_{3}},\\ & & { \mbox{$[{\bf k}^{(q)2},z^{jlsi}_{kk_{3}}]$ } } = k(k+1 ) z^{jlsi}_{kk_{3}},\\ & & { \mbox{$[{\bf j}^{(q)2},z^{jlsi}_{kk_{3}}]$ } } = j(j+1 ) z^{jlsi}_{kk_{3}},\\ & & { \mbox{$[{\bf l}^{(q)2},z^{jlsi}_{kk_{3}}]$ } } = l(l+1 ) z^{jlsi}_{kk_{3}},\\ & & { \mbox{$[{\bf s}^{(q)2},z^{jlsi}_{kk_{3}}]$ } } = s(s+1 ) z^{jlsi}_{kk_{3}},\\ & & { \mbox{$[{\bf i}^{(q)2},z^{jlsi}_{kk_{3}}]$ } } = i(i+1 ) z^{jlsi}_{kk_{3}}.\end{aligned}\ ] ] then the chiral fields can be expanded in the series of @xmath89 : @xmath90 equation ( [ eq : k3_symmetry ] ) means that eq .
( [ eq : mode_u ] ) contain only @xmath91 components .
similarly , the quark fields are expanded in the kana - ripka bases @xmath92 @xcite : @xmath93 where the subscript @xmath94 indicates quantum numbers other than @xmath95 and @xmath82 .
the matrix elements of @xmath65 with @xmath92 have a contribution only from the bases with the same @xmath82 .
the lowest order perturbative expansion of the action @xmath68 is given by @xmath96 where @xmath97 is a sufficiently large time interval and @xmath98 is the classical soliton energy . with the eigenvalues @xmath99 of @xmath65 and @xmath100 of @xmath101
we obtain @xmath102 \epsilon_{n } - \sum_{n } \rho^{vac}_{\lambda}(\epsilon^{0}_{n } ) \epsilon^{0}_{n},\ ] ] where @xmath103 is the occupation number of the valence quark level @xmath99 and @xmath104 is the cutoff function of the vacuum energy for level @xmath99 with cutoff parameter @xmath2 ( appendix ) . in this article
, we use schwinger s proper time regularization @xcite and set @xmath105 mev , and we assume that the valence quark is in the lowest positive energy state . to subtract @xmath98 , we use @xmath106 but @xmath107 the reason is the following .
physically , @xmath98 is the static soliton energy in the body fixed frame and its subtraction point is the dirac sea in the absence of a static soliton .
we employ the body fixed frame in which @xmath53 and @xmath54 are nonzero . in this frame
the hamiltonian without the static soliton is @xmath106 .
the equations of motion for the profile functions @xcite are obtained by the extremum conditions for the action @xmath108 with respect to the radial component @xmath28 and the direction of the unit vector @xmath29 : @xmath109 \frac{{\bf p}({\bf r})}{{\mbox{$\left| { \bf p}({\bf r } ) \right|$}}}. \label{eq : eom_angle}\end{aligned}\ ] ] here , @xmath110 and @xmath111 are the scalar - isoscalar and pseudoscalar - isovector densities , respectively , and are defined by @xmath112 where @xmath113 , @xmath114 is the eigenvector belonging to the eigenvalue @xmath99 of @xmath65 , and @xmath115 is the cutoff function shown in the appendix . using the boundary conditions @xmath116 eqs .
( [ eq : eom_f]),([eq : eom_angle ] ) and the dirac equation with the hamiltonian eq .
( [ eq : h0 ] ) are self - consistently solved @xcite .
both the profile functions and the classical soliton energy are some even functions of @xmath54 because of the isospin symmetry of the model .
the value of @xmath53 mainly affects the @xmath83 dependence of the profile functions and the value of @xmath54 affects the @xmath84 dependence of the profile functions .
it was found from the calculation that @xmath28 hardly depends on @xmath117 .
thus , we can treat @xmath28 as a function of @xmath83 only .
the profile @xmath118 is shown for the cases of @xmath119 with @xmath120 in fig . [
fig : profile1 ] .
it is shown below that the range @xmath121 corresponds to the octet baryon .
on the other hand , the unit vector @xmath29 has axial symmetry about the @xmath81 axis and plane symmetry with respect to the @xmath122 plane .
the deviation from the hedgehog shape is an increasing function of @xmath117 and takes a maximum at @xmath123 and a minimum at @xmath124 in real space .
we show @xmath29 at @xmath125 in fig .
[ fig : profile2 ] . with @xmath120 .
the solid line represents the function at @xmath126 and the dashed line at @xmath127 . ] in isospin space with @xmath128 and @xmath129 .
the solid arrow represents the vector at @xmath126 and the dashed arrow at @xmath127 . ] .
the range of @xmath54 is restricted to @xmath130 . ]
figure [ fig : ecl_2d ] displays the @xmath54 dependence of @xmath98 . each curve corresponds to a value of @xmath53 , and the range of @xmath54 is restricted by eq .
( [ eq : ddtrange ] ) .
the classical soliton energy @xmath98 is an increasing function of @xmath53 around @xmath127 and a decreasing function of @xmath117 at a fixed @xmath53 .
because the deviation from the hedgehog shape increases with @xmath117 , the soliton takes a stable deformed shape at @xmath131 .
there are two expansion parameters for the effective action @xmath132 : the number of colors @xmath0 and the su(3 ) symmetry breaking @xmath43 .
we expand @xmath132 up to second order in powers of @xmath133 and first order in powers of @xmath43 .
first , we define the local variables @xmath94 , @xmath134 , and @xmath135 of the rotation @xmath136 , @xmath32 , and @xmath48 , respectively , and the fluctuation @xmath137 by @xmath138 where @xmath139 , @xmath140 , and @xmath141 .
since in the large @xmath0 limit @xcite , @xmath142 we expand @xmath132 up to second order in @xmath143 and @xmath144 , and first order in @xmath137 .
thus , we get the lagrangian @xcite @xmath145 where the coefficients are defined as @xmath146 here , @xmath147 and @xmath148 are the cutoff functions shown in the appendix . the indices @xmath149 and @xmath150 denote the origin of the vacuum part of the coefficients from the real and imaginary parts of the action @xmath132 in the imaginary time prescription .
we subtract the vacuum contribution of the eigenstates of @xmath106[eq .
( [ eq : kinetic ] ) ] from these quantities in a similar way to @xmath98 . by @xmath82 symmetry ,
( [ eq : leff_g ] ) becomes @xmath151 since @xmath143 and @xmath137 in eq .
( [ eq : leff_g ] ) are functions of @xmath50,@xmath51 and @xmath152 , we should further expand these quantities in the actual calculation .
next , we would like to express the hamiltonian in terms of @xmath134,@xmath135,@xmath50 and their canonically conjugate momenta .
the canonical momenta conjugate to these coordinates are defined by @xmath153 and satisfy the following commutation relations : @xmath154 $ } } & = & { \mbox{$[j_{i},b^{j}]$ } } = \frac{1}{i } \delta_{i}^{j},\\ { \mbox{$[p^{\gamma},d^{\dagger}_{\beta}]$ } } & = & { \mbox{$[p^{\dagger}_{\beta},d^{\gamma}]$ } } = \frac{1}{i } \delta^{\gamma}_{\beta}.\end{aligned}\ ] ] the collective hamiltonian derived from eq .
( [ eq : leff ] ) , @xmath155 is calculated up to order @xmath133 .
because of the @xmath82 symmetry of the chiral fields , there is the following constraint on the canonical momenta : @xmath156 here @xmath157 is the third component of the isospin carried by the kaon , @xmath158 if we assume the hedgehog shape for the chiral fields , the constraint becomes @xmath159
before diagonalizing @xmath160 [ eq .
( [ eq : heff ] ) ] , we will describe @xmath161 , which is the @xmath162 hamiltonian of @xmath160 in the large @xmath0 limit in the mean field approximation : @xmath163 d^{\dagger } d + \sqrt{3 } \left(\frac{b_{8}}{4 \phi_{(+ ) } } + m_{8 } \delta_{(+)}\right ) i \left(d^{\dagger}p - p^{\dagger}d\right ) \nonumber\\ & & + 3 \left [ \frac{b_{8}^{2}}{4 \phi_{(- ) } } + 2 m_{8 } \left(\delta_{(- ) } b_{8}-\frac{\gamma_{3}}{\sqrt{3}}\right ) \right ] d^{\dagger } \tau_{3 } d + \sqrt{3 } \left(\frac{b_{8}}{4 \phi_{(- ) } } + m_{8 } \delta_{(-)}\right ) i \left(d^{\dagger}\tau_{3}p - p^{\dagger}\tau_{3}d\right),\nonumber\\ \label{eq : h_n0}\end{aligned}\ ] ] where @xmath164 the eigenstate of @xmath161 describes the kaon in the background soliton .
the energy @xmath165 is induced by the deviation to the strange and isospin direction from the su(2 ) hedgehog soliton . here the @xmath166 term in @xmath165 gives a negative contribution to @xmath161 and @xmath167 a spositive one .
the hamiltonian @xmath161 is in a bilinear form of the isodoublets @xmath50 and @xmath168 , and the type of separation of variables with respect to the individual component . for the individual components , the hamiltonian except for the constant terms
is given by @xmath169 where @xmath170 are constants for each component .
if @xmath170 satisfy @xmath171 @xmath161 can be diagonalized exactly using the following transformations @xcite : @xmath172 where @xmath173 and @xmath174 are constants depending on @xmath170 .
the quantities @xmath175 ( @xmath176 ) and @xmath177 ( @xmath178 ) are the creation ( annihilation ) operators for states with the same quantum numbers as the kaon and antikaon , respectively .
these satisfy the commutation relations @xmath179 $ } } & = & { \mbox{$[\bar{\xi}_{\gamma},\bar{\xi}_{\beta}^{\dagger}]$ } } = \delta_{\gamma\beta},\\ { \mbox{$[\xi_{\gamma},\bar{\xi}_{\beta}]$ } } & = & { \mbox{$[\xi_{\gamma},\bar{\xi}_{\beta}^{\dagger}]$ } } = 0.\end{aligned}\ ] ] this then finally leads to the diagonalized form @xmath180 where @xmath181 there are also two physical quantities diagonalized by the creation and annihilation operators : the third component of the isospin carried by the kaon [ eq . ( [ eq : isospin_s ] ) ] , @xmath182 and the strangeness carried by the kaon , @xmath183 the fock space is generated by successive operation of the creation operators on the vacuum state @xmath184 : @xmath185 where @xmath186 and @xmath187 are some positive integers . in the mean field approximation , @xmath53 and @xmath54
should be self - consistently determined by eqs .
( [ eq : ddt0 ] ) and ( [ eq : ddt3 ] )
. then the stability of the approximation should be checked . because the coefficients of the individual terms of @xmath161 are evaluated using a soliton depending on @xmath53 and @xmath54 ,
the potential term is physically meaningful only in the vicinity of the expectation values .
it is difficult to draw a potential diagram over a wide range of @xmath50 and @xmath51 .
the reason is that @xmath161 describes the combined system of the classical soliton and the kaon . in order to investigate the stability of the mean field approximation we will treat @xmath53 and @xmath54 as parameters at first . when @xmath53 and @xmath54 are different from the expectation values of eqs .
( [ eq : ddt0 ] ) and ( [ eq : ddt3 ] ) , the differences @xmath188 act on the system as a kind of external field .
therefore , by investigating the @xmath53 and @xmath54 dependencies of the lowest eigenvalue @xmath189 of @xmath161 , we can study the stability of the system against external perturbation .
figure [ fig : mean_field ] shows the behavior of @xmath189 as a function of @xmath53 and @xmath54 in the cases of @xmath190 .
@xmath189 is an even function of @xmath54 like @xmath98 . , where @xmath53 and @xmath54 are treated as parameters .
] we further investigate the characteristic behavior of @xmath189 in the three cases .
figure [ fig : e_n0_dd0 ] shows the @xmath53 dependence of @xmath189 .
for @xmath191 , the @xmath53 dependence of @xmath189 is similar to that of the su(2 ) sector quark mass @xmath192 .
it is due to the @xmath166 term in @xmath165 of eq .
( [ eq : h_n0 ] ) that @xmath189 is flat compared with @xmath98 or @xmath192 . in the case of @xmath193 , @xmath189 shows a similar tendency to @xmath194 . in the case of @xmath195
, @xmath189 is similar to @xmath196 .
figure [ fig : e_n0_dd3 ] shows the @xmath54 dependence of @xmath189 .
it is due to @xmath167 term in @xmath165 that @xmath189 is convex downward as a function of @xmath54 for @xmath191 . for @xmath197 , since @xmath165 is canceled by the potential of @xmath50 and @xmath51 , @xmath189 is convex upward due to the @xmath54 dependence of @xmath198 ( fig .
[ fig : meff ] ) . and the classical soliton energy in the cases of @xmath120 and @xmath128 . between the two curves with @xmath191 ,
the case of @xmath120 corresponds to lower energy .
for @xmath197 it corresponds to higher energy . ] and the classical soliton energy at @xmath199 . ] therefore , in the case of @xmath191 it is energetically forbidden that the soliton deviate from the hedgehog shape , and the harmonic analysis around @xmath200 is justified . on the other hand , in cases of @xmath197 @xmath189
is energetically unstable at @xmath60 .
thus the soliton deviates from the hedgehog shape . however , it is necessary to keep in mind that under the condition of eq .
( [ eq : hcoff_cond ] ) the potential term of @xmath161 is convex downward and instability for @xmath50,@xmath51 is not caused .
thus , the instability is due to the classical soliton .
if the degrees of freedom for the @xmath54 direction are not included in the soliton , the instability does not occur . in table
[ tb : mean_field ] we show @xmath53 , and @xmath201 , which are self - consistently determined in the mean field approximation .
.@xmath53 , , and @xmath202 for @xmath190 in the mean field approximation . [ cols="^,^,^,^,^",options="header " , ] [ tb : particle ] ) .
the superscript @xmath203 of @xmath204 represent the results of the hedgehog soliton .
denotes the experimental values . ] in the eigenstates of every baryon , there are large transitions among the bases with different values of @xmath205 , that is , remains of the clebsch - gordan series for the constraint eq .
( [ eq : constraint ] ) which is maintained only in the hedgehog shape . in the case of @xmath1 ,
the energy eigenvalues are increased by the transition .
the situation does not change on incorporating the @xmath206 states . for @xmath2 and @xmath3 ,
an interaction induced by the deformation of the soliton picks up the transition .
then the energy shows a decrease . that is caused only in the space with @xmath207 . for @xmath4
, there is a similar decrease in the space that takes account of @xmath206 perturbatively .
the contributions of @xmath206 to the energy take negative values for every baryon due to the second order perturbation formula .
the stability of the hedgehog shape has been investigated for the octet baryon by means of the chiral quark soliton model . in the mean field approximation
, it was shown that the stable form of the soliton changes according to the strangeness of the baryon . in the case of the soliton without strangeness ,
the hedgehog shape is stable . on the other hand , for the soliton with strangeness
an instability occurs not only in the strange direction but also in the isotopic one .
the instability has a global nature due to the inertial force , and one can not find it from study of the curvature of the local potential .
it is necessary to incorporate the degrees of freedom for the deviation in the ansatz of the soliton .
after the collective quantization has been performed , the states identified as the strange baryons ( @xmath2,@xmath3,@xmath4 ) also energetically favor the deformed soliton .
the strange baryon masses are in good agreement with both the absolute value and the difference among the experimental values .
these become small compared with the results of the mean field approximation due to the interaction caused by the deformation of the soliton . on the other hand
, the approach with the hedgehog soliton reproduces the mass difference among the baryons well but the absolute value . also , the collective states have higher energy than the classical soliton .
weigel _ et al .
_ investigated the quantum correction due to the zero modes off the hedgehog soliton in the njl model @xcite .
the correction gives a large negative contribution to the @xmath1 and @xmath208 masses , c and the results are in good agreement with the experimental values . in our approach ,
the mass of @xmath1 is somewhat too large due to the hedgehog shape of the soliton .
therefore , both approaches cooperatively would be in agreement with the experimental value for @xmath1 .
on the other hand , the deformed soliton corresponding to the strange baryon has less symmetry than the hedgehog soliton . because of the estimation @xcite relied on the hedgehog shape @xcite , the relation between the two approaches is obscure at present .
the eigenvalue of the hamiltonian up to order 1 in the large @xmath0 limit is an even function of @xmath54 .
therefore the mean field energy at @xmath54 is degenerate with that at @xmath209 .
a more accurate calculation of the fluctuation off the mean field should treat the tunnel effect .
it is in progress to perform the description of the decuplet baryon and to incorporate the local variation of the kaon wave function .
we would like to thank m. kawabata and h. kondo for many fruitful discussions of this problem .
the cutoff function of the vacuum energy for the energy level @xmath99 is given by @xmath210 with the cutoff parameter @xmath2 , where , using the schwinger proper time regularization , @xmath211 may be written as @xmath212 the cutoff function for first order matrix element is given by @xmath213 with @xmath214 the indices @xmath149 and @xmath150 denote the origin of the function from the real and imaginary parts of the action @xmath132 in the imaginary time prescription .
similarly , for the second order matrix elements one obtain @xmath215 where @xmath216 are given by @xmath217 99 g. t hooft , nucl . phys .
* b72 * , 461 ( 1974 ) ; * b75 * , 461 ( 1974 ) .
e. witten , nucl .
* b160 * , 57 ( 1979 ) .
r. dashen , e. jenkins , and a. v. manohar , phys .
d * 49 * , 4713 ( 1994 ) .
t. h. r. skyrme , proc .
r. soc .
london * a260 * , 127 ( 1961 ) . g. adkin , c. nappi , and e. witten , nucl .
* b228 * , 552 ( 1983 ) .
a. p. balachandran , f. lizzi , and v. g. j. rodgers , nucl .
* b256 * , 525 ( 1985 ) .
e. guadanini , nucl .
* b236 * , 35 ( 1984 ) .
p. o. mazur , m. a. nowak , and m. praszalowicz , phys .
lett . * 147b * , 137,(1984 ) .
h. yabu and k. ando , nucl . phys .
* b301 * , 601 ( 1988 ) .
y. nambu and g. jona - lasinio , phys .
* 122 * , 345 ( 1961 ) ; * 124 * , 246 ( 1961 ) .
d. ebert and h. reinhardt , nucl .
b271 * , 188 ( 1986 ) . h. weigel , r. alkofer , and h. reinhardt , nucl .
* b387 * , 638 ( 1992 ) .
d. i. diakonov , v. yu . petrov , and p. v. pobylitsa , nucl .
* b306 * , 809 ( 1988 ) . h. reinhardt and r. wnsch , phys .
lett . b * 215 * , 577 ( 1988 ) . h. reinhardt , nucl .
* a503 * , 825 ( 1989 ) . m. wakamatsu and h. yoshiki , nucl .
* a524 * , 561 ( 1991 ) . m. wakamatsu and n. kaya , prog . theor .
* 95 * , 767 ( 1996 ) .
v. christov , a. blotz , h. c. kim , p.
v. pobylitsa , t. watabe , th .
meissner , e. ruiz arriola , and k. goeke , prog .
. phys . * 37 * , 91 ( 1996 ) .
r. alkofer , h. reinhardt , and h. weigel , phys . rep .
* 265 * , 139 ( 1996 ) .
d. i. diakonov , v. yu . petrov , p. v. pobylitsa , m. v. polyakov , and c. weiss , nucl .
phys . * b480 * , 341 ( 1996 ) . h. weigel , l. gamberg , and h. reinhardt , phys . lett .
b * 399 * , 287 ( 1997 ) . m. wakamatsu and t. kubota , phys .
d * 57 * , 5755 ( 1998 ) .
h. weigel , h. reinhardt , and r. alkofer , phys .
* b313 * , 377 ( 1993 ) .
a. chodos and c. b. thorn , phys .
d * 12 * , 2733 ( 1975 ) .
a. hosaka and h. toki , _ quarks , baryons and chiral symmetry _ ( world scientific , singapore , 2001 ) .
m. takayama , a. hosaka , and h. toki , prog .
phys . * 101 * , 1271 , 1999 a. blotz , d. i. diakonov , k. goeke , n. w. park , v. yu . petrov , and p. v. pobylitsa , nucl .
* a555 * , 765 ( 1993 ) .
e. braaten and l. carson , phys .
d * 38 * , 3525 ( 1988 ) .
d. kaplan and i. klebanov , nucl .
* b335 * , 45 ( 1990 ) . c. callan and i. r. klebanov , nucl .
b262 * , 365 , ( 1985 ) . y. oh , b. y. park , and d. p. min , phys .
d * 49 * , 4649 ( 1994 ) .
s. kahana and g. ripka , nucl . phys . * a429 * , 462 ( 1984 ) .
j. schwinger , phys .
rev . * 82 * , 664 ( 1951 ) .
k. m. westerberg and i. r. klebanov , phys .
d * 50 * , 5834 ( 1994 ) .
h. weigel , r. alkofer , and h. reinhardt , nucl .
* a582 * , 484 ( 1995 ) .
r. alkofer and h. reinhardt , _ chiral quark dynamics _ , springer - verlag , berlin , 1995 . | in this paper the stability of the hedgehog shape of the chiral soliton is studied for the octet baryon with the su(3 ) chiral quark soliton model .
the strangeness degrees of freedom are treated by a simplified bound - state approach , which omits the locality of the kaon wave function .
the mean field approximation for the flavor rotation is applied to the model .
the classical soliton changes shape according to the strangeness .
the baryon appears as a rotational band of the combined system of the deformed soliton and the kaon . |
the muon - spin rotation ( @xmath0sr ) technique is a powerful tool to study the local magnetic field distribution in solids.@xcite it is also successfully employed to gain valuable information on type - ii superconductors by probing the magnetic field distribution @xmath6 generated by a vortex lattice ( for a review mostly regarding cuprate high - temperature superconductors , see , e.g. ref . ) .
if the vortex - lattice configuration is ordered and static , from the obtained @xmath6 the characteristic length scales of the superconductor like the magnetic penetration depth @xmath7 or with reservations the ginzburg - landau coherence length @xmath8 can be extracted in a reliable way . however , the vortex arrangement in cuprate superconductors is commonly not an ideal static two - dimensional hexagonal flux - line lattice .
individual vortices are always subject to displacements due to pinning , e.g. at grain boundaries or lattice defects . also either thermal or quantum fluctuations introduce dynamics and may cause a reordering and relaxation of a non - equilibrium vortex lattice .
reviews on this subject can be found , e.g. in refs . and .
the presence of those imperfections changes the @xmath0sr field distributions : while weak random pinning leads to a symmetric broadening of @xmath6,@xcite thermal fluctuations might even change the observed asymmetry of the lineshape as seen in the example of the vortex - lattice melting in the bi@xmath9sr@xmath9cacu@xmath9o@xmath10 compound.@xcite moreover , the lineshape asymmetry has also been found to depend on three - body correlations between the vortices as seen in the so - called `` vortex - glass state '' in la@xmath11sr@xmath12cuo@xmath3.@xcite + therefore , in order to obtain accurate information on the superconducting - state parameters , it is preferable to complement transverse - field ( tf ) @xmath0sr measurements in the mixed state by investigations using other experimental techniques . here
, we report on combined tf @xmath0sr and magnetization studies of the mixed state of a la@xmath1sr@xmath2cuo@xmath3 single crystal in a low magnetic field of @xmath13 mt applied parallel to the @xmath5 axis of the crystal .
a detailed analysis shows that although the measurements are not susceptible to small variations in the order - parameter symmetry , the data can be overall consistently described by taking into account a single energy gap in the quasiparticle excitation spectrum with @xmath14 symmetry and a change in the sampling of the spatial field distributions by the muons due to vortex - pinning effects .
a cylindrical la@xmath1sr@xmath2cuo@xmath3 single crystal with a diameter of @xmath15 mm and a height of @xmath16 mm was used for the @xmath0sr studies .
the crystal was grown by the travelling - solvent floating - zone technique similar to the one reported in ref . .
the subsequent characterization by laue x - ray diffraction showed the @xmath5 axis pointing perpendicular to the cylinder axis .
x - ray powder diffraction on a sample of the same batch indicated a single crystalline phase .
the magnetization studies have been carried out in a quantum design 5 t mpms squid magnetometer on a small piece of approximate dimensions @xmath17
mm@xmath18 of the same batch with the @xmath5 axis along the shortest edge of the sample .
the critical temperature @xmath19 is about @xmath20 k ( cf . fig .
[ mvstfc ] ) .
the magnetization measurements in a magnetic field in the range from a few millitesla up to several tesla applied parallel to the @xmath5 axis of the crystal were performed using different field - cooling ( fc ) procedures : ( i ) fast cooling from above @xmath19 to @xmath21 k with a high rate of @xmath22 k / min and measuring during warming up the sample ( @xmath23 ) , ( ii ) slow cooling from above @xmath19 to @xmath21 k with a low rate of @xmath24 k / min and measuring during this cooling ( @xmath25 ) , and ( iii ) measuring during warming up from the slowly field - cooled state obtained through ( ii ) ( @xmath26 ) .
+ the @xmath0sr experiments with magnetic fields up to @xmath4 mt applied along the @xmath5 axis were performed at the gps spectrometer at the @xmath27m3 beam line at the swiss muon source at the paul scherrer institute.@xcite the used helium flow cryostat allows to cool the sample from @xmath28 k to @xmath29 k in about two minutes at the maximum cooling rate .
it is therefore possible to do the @xmath0sr measurements using equivalent cooling procedures as for the magnetization studies .
it should also be noted that the ratio of the sample dimensions parallel and perpendicular to the magnetic field has been similar for the magnetization and the @xmath0sr measurements .
thus , the demagnetization effects on the respective samples are similar for both studies and overall small due to the small absolute value of the susceptibility in the field - cooled mixed state .
+ for the @xmath0sr experiments spin - polarized positively charged muons with an energy of @xmath30 mev are implanted into the sample and thermalize .
the spin ensemble interacts with its local environment until the muons decay ( @xmath31s ) and emit the formed positrons preferentially in the direction of the muon spin at the time of decay .
thus , by detecting the muons at their implantation time and the positrons after the decay the temporal evolution of the muon - spin polarization in a sample may be recorded and by that information about the local environment of the muons is obtained . in a static local magnetic field
@xmath32 with a non - zero component perpendicular to the muon spins , these undergo a larmor precession with a frequency @xmath33 , where @xmath34 mhz / t is the gyromagnetic ratio of the muon . in a homogeneous transverse field
the measured field distribution would therefore ideally be a dirac @xmath35 function , whereas @xmath6 has a finite width when spatially inhomogeneous fields are probed by the muons . in case
the local fields within the mixed state of a type - ii superconductor are sampled , this leads to a @xmath7- and @xmath8-dependent characteristic asymmetric @xmath6 with a `` high - field tail '' originating from contributions of muons stopping in and close to the vortex cores.@xcite further details concerning the analysis of these field distributions are introduced in sec .
[ sec : musrdata ] ; more information on @xmath0sr techniques in general can be found in ref . .
the magnetization measurements in an applied magnetic field of @xmath4 mt are presented in fig .
[ mvstfc ] .
the overall diamagnetic signals exhibit certain peculiarities .
while the slowly cooled @xmath26 data represent the magnetization curve of a rather equilibrated mixed state , the fast cooled @xmath23 measurement shows an initial low - temperature magnetization which deviates from the @xmath26 curve , reflecting a fairly undefined vortex configuration obtained by this cooling procedure .
in contrast , the @xmath25 data show a hump below @xmath39 k. sr@xmath2cuo@xmath3 for @xmath40 mt and @xmath41 for the different cooling procedures described in the text . ]
the field - cooled magnetization of type - ii superconductors during cooling ( fcc ) and during warming ( fcw ) were examined theoretically by clem and hao.@xcite they demonstrated that the important parameters in analyzing the irreversible behavior of the low - field dc susceptibilities are the critical current density , the sample dimensions , and the lower critical field @xmath42 .
however , the theoretical model developed in ref . is subject to various simplifications and can not be directly adapted to the curves presented in fig .
[ mvstfc ] . nevertheless , the detected hysteresis between the cooling and warming measurements
is readily explained by differences in the flux expulsion and reentering in the presence of vortex pinning.@xcite in order to elucidate the origin of the distinct observations in the case that the sample is cooled very fast , the magnetization relaxation of the @xmath23 and @xmath25 mixed states was studied over a time of about six hours for various temperatures between @xmath21 k and @xmath19 as depicted in fig .
[ mrelaxation-20mt ] .
it is seen that for temperatures below about @xmath16 k the vortex configuration hardly relaxes at all . for higher temperatures
the magnetization returns to the `` equilibrium one '' with a maximum in the relaxation rate between about @xmath4 k and @xmath43 k which decreases again for temperatures above .
figure [ relrates - vs - t ] shows the relaxation rate , defined as the slope of the measured magnetic moment in fig .
[ mrelaxation-20mt ] for times @xmath44 s. the vortex relaxation for @xmath23 and @xmath25 is similar for both studied situations . the small observed differences in the relaxation rates seem to be merely related to the differing initial conditions .
even though a fully quantitative analysis of the vortex relaxation is beyond the scope of this article , the low - field flux - line dynamics appears to be different in at least the three temperature regions described before .
a similar behavior is found in bi@xmath9sr@xmath9cacu@xmath9o@xmath10 where especially in a low magnetic field different vortex - pinning regimes have been identified.@xcite yet , la@xmath1sr@xmath2cuo@xmath3 is far less anisotropic than bi@xmath9sr@xmath9cacu@xmath9o@xmath10 and the more three - dimensional vortex structure is dominated by josephson coupling rather than electromagnetic coupling .
thus , the pinning of individual `` vortex pancakes '' like in bi@xmath9sr@xmath9cacu@xmath9o@xmath10 at low temperatures appears unlikely .
however , the overall observed magnetization relaxation still suggests a crossover between distinct vortex - pinning scenarios also in la@xmath1sr@xmath2cuo@xmath3 .
+ it should be noted as well that the applied field of only @xmath4 mt at low temperatures is _ smaller _ than @xmath45 in almost optimally doped la@xmath1sr@xmath2cuo@xmath3.@xcite as can be seen in fig .
[ mvstfc ] this does not affect the macroscopic `` equilibrium magnetization '' , yet , the complete loss of dynamics for @xmath46 k might reflect the `` final freeze - in '' of the flux distribution at about the temperature where @xmath47 mt.@xcite + mt in the mixed state of la@xmath1sr@xmath2cuo@xmath3 obtained for the different cooling procedures .
@xmath48 denotes the magnetic moment at @xmath49 defined for each temperature by data similar to those shown in fig .
[ mvstfc ] . ]
sr@xmath2cuo@xmath3 in @xmath40 mt as a function of temperature for the two studied cooling procedures .
the rates are determined from the data shown in fig .
[ mrelaxation-20mt ] for @xmath44 s. ] similar differences between fcc and fcw magnetization data have also been observed for other applied fields ( with slightly shifted temperature intervals ) .
however , previous @xmath0sr experiments on a single crystal of la@xmath1sr@xmath2cuo@xmath3@xcite have shown that the field of @xmath4 mt is high enough to introduce a sufficient amount of vortices in the material which generate the characteristic magnetic - field distribution in the type - ii superconductor .
therefore , the magnitude of the applied field has also been chosen to be @xmath4 mt for the present study . in the following it
shall be investigated how the @xmath0sr results change for the various vortex configurations studied above .
for the initial @xmath0sr studies the sample was cooled from above @xmath19 to the lowest temperature in the applied field @xmath51 at a cooling rate of @xmath52 k / min and the data were recorded while warming up the sample stepwise to @xmath53 .
this temperature cycle corresponds to the @xmath23 procedure of the magnetization measurements .
first , the resulting data were analyzed using a sum of three oscillating signals with gaussian damping ( `` 3-gaussian '' method ) which has been shown to be a reasonable approximation of the data if the probed field distribution originates from an ordered ( or weakly distorted ) vortex lattice.@xcite in this case , the muon - decay asymmetry which is proportional to the spin polarization is modeled by @xmath54 here the @xmath55 are the partial asymmetries of the three signal contributions , @xmath56 the respective gaussian depolarization rates , and @xmath57 the average fields of each of the fractions .
the @xmath55 are common parameters for all temperatures whereas the @xmath56 and @xmath57 are temperature - dependent .
@xmath58 is the common initial phase of the muon spins with respect to the positron detector .
the total decay asymmetry is @xmath59 and the second central moment of the corresponding field distribution is given by@xcite @xmath60 , \label{second - moment}\ ] ] where @xmath61 is the first moment of the field distribution . assuming the depolarization of the muon - spin ensemble is caused only by the inhomogeneous field distribution generated by the vortex lattice and random nuclear moments , then the contribution of the flux - line lattice to the second central moment is given by @xmath62 , where @xmath63 is the muon - spin depolarization rate above @xmath64 k in the normal state of the material . in order to further characterize the obtained field distributions
also the dimensionless skewness parameter @xmath65 is calculated ; it represents the lineshape asymmetry of @xmath6.@xcite given the sum of gaussian distributions , the third central moment can be written as @xmath66 .
\label{third - moment}\ ] ] + for the final interpretations within the above model , the assumption@xcite @xmath67 ( @xmath68 is the in - plane magnetic penetration depth ) is essential . in order to test this assumption the data were also analyzed using an analytic ginzburg - landau ( agl ) model.@xcite here the spatial distribution of the magnetic field in the vortex state of a superconductor is modeled by the fourier series @xmath69}{\left(f_{\infty}^2 + k^2\lambda_{ab}^2\right)^{\frac{1}{2}}k_1\left(\dfrac{\xi_{\mathrm{v}}}{\lambda_{ab}}f_{\infty}\right)}\exp\left(-\imath{\mathbf{k}}{\mathbf{r}}\right ) , \label{agl - model}\ ] ] where @xmath70 , @xmath71 are the reciprocal lattice vectors of a two - dimensional hexagonal vortex lattice , @xmath72 , and @xmath73 is a modified bessel function of the second kind .
the parameter @xmath74 representing the suppression of the superconducting order parameter due to overlapping vortex cores has been set equal to @xmath75 ( no suppression ) in the analysis since the vortices are far apart in the low applied field used .
the length @xmath76 is the _ effective _ vortex - core radius which is a variable parameter within this model .
the field distribution @xmath6 probed by the muons is obtained by random sampling of @xmath77 and finally the muon - decay asymmetry is given by @xmath78 the gaussian prefactor in eq .
( [ asy - from - pb ] ) accounts for broadening of the field distribution by nuclear dipole fields as well as weak random pinning.@xcite for the present data @xmath79 is growing continuously from @xmath80 just below @xmath19 to @xmath81 at @xmath29 k ( cf . fig .
[ fch]d ) .
+ sr data on la@xmath1sr@xmath2cuo@xmath3 : temperature dependence of ( a ) @xmath82 obtained from the agl model and @xmath83 from the `` 3-gaussian '' analysis ; the solid red line is a fit to the semi - classical model of ref .
using an order parameter with @xmath14 symmetry , ( b ) the probed mean field , ( c ) the skewness parameter @xmath84 , ( d ) the width of the gaussian in eq .
( [ asy - from - pb ] ) accounting for symmetric broadening of @xmath6 within the agl model and the effective vortex - core radius @xmath76 , and ( e ) the inverse width of the lorentzian weighting function in eq .
( [ lorentzian ] ) given in units of the inverse inter - vortex distance of a fully ordered hexagonal flux - line lattice .
@xmath85 , where @xmath86 is the flux quantum . for further details ,
see text . ]
the comparison of @xmath83 obtained by the `` 3-gaussian '' analysis ( black squares in fig .
[ fch]a ) with @xmath82 as determined by the agl analysis ( blue diamonds in fig .
[ fch]a ) reveals that for this set of data both parameters can not be simply scaled to match in the full temperature range , yet , both curves show pronounced changes in their curvatures around @xmath87 k. the skewness parameter @xmath84 is depicted in fig . [ fch]c . below @xmath88
k @xmath84 is virtually constant ; the small differences in the absolute value of @xmath84 between the various models are merely the result of effectively cutting off high - field contributions in the `` 3-gaussian '' analysis , thus leading to an overall smaller @xmath84 which is less susceptible to minor effects in the `` high - field tail '' of @xmath6 .
one may note that the `` 3-gaussian '' analysis exhibits a strong drop and a sign change in @xmath84 at @xmath89 k. while such behavior usually is attributed to a melting of the vortex lattice,@xcite here the detailed temperature dependence of @xmath84 is most likely an artifact of the fitting procedure involving three gaussians for an almost symmetric field distribution slightly below @xmath19 .
this view is supported by the fact that the irreversibility line in la@xmath90sr@xmath91cuo@xmath3 at this low applied magnetic field tends to be much closer to @xmath19.@xcite + however , it is most important to check the temperature dependence of @xmath92 determined by the different analyses .
figure [ fch]b shows @xmath92 for the above described models as well as its values obtained by a maximum - entropy approach@xcite which has the advantage that it is _ not _ tied to any model . for the maximum - entropy analysis an apodization with a gaussian ( @xmath93 )
was applied to the asymmetry spectra .
this leads to additional symmetric broadening of the determined field distribution but does not change @xmath92 .
the various analyses all yield qualitatively the same results : while the mean field is slightly diamagnetically shifted and approximately constant for @xmath94 k , for @xmath95 k @xmath92 strongly _ rises _ with decreasing temperature even substantially above the applied field . additionally , as shown in fig .
[ fch]d , in order to explain the data @xmath76 obtained from the agl fit would have to increase drastically in the same temperature range ( cf .
[ fch]d ) whereas it assumes values between about @xmath16 nm and @xmath96 nm above @xmath87 k before it diverges close to @xmath19 .
+ since the magnetization measurements have revealed a change in the vortex dynamics at @xmath97 k and especially that the mixed state , generated in this initial @xmath0sr experiment , is far from being equilibrated it shall further be investigated if the increase in @xmath92 , @xmath76 , and the second central moment of the @xmath0sr field distributions at low temperatures is related to those observations . for this purpose
, additional @xmath0sr measurements have been performed . in a first step
the single crystal has been cooled slowly in the applied magnetic field from above @xmath19 to finally @xmath29 k with a cooling rate of @xmath98 k / min while the measurements have been done at intermediate stable temperature steps lasting about one hour each ( @xmath25 ) .
subsequently , the so generated `` equilibrium vortex state '' served as starting point for another series of measurements which have been conducted during the stepwise warming to above the critical temperature ( @xmath26 ) .
sr data on la@xmath1sr@xmath2cuo@xmath3 for the different cooling procedures .
the insets show the corresponding asymmetry spectra for @xmath99 k ( black squares ) and @xmath29 k ( orange circles ) , respectively .
the solid lines in the insets are fits to the data using eq .
( [ asy - from - pb ] ) . for @xmath29
k , @xmath6 is calculated using eqs .
( [ agl - model ] ) and ( [ lorentzian ] ) ; for @xmath100 , @xmath101 . ]
tf @xmath0sr measurements of la@xmath1sr@xmath2cuo@xmath3analogous to fig .
[ fch ] . ]
tf @xmath0sr measurements of la@xmath1sr@xmath2cuo@xmath3analogous to fig .
[ fch ] . ]
figure [ maxent ] shows a comparison between the static field distributions obtained by the maximum - entropy analysis of the tf @xmath0sr data as well as selected asymmetry spectra for the different cooling procedures the differences are only marginal .
the further analysis with the above described models confirms this finding ( cf . figs .
[ scc]a d and [ sch]a d ) . also , the determined field distributions presented in fig .
[ maxent ] agree well with the expectations where , e.g. given the model parameters obtained by the agl analysis at the lowest temperature the `` high - field tail '' of @xmath6 should range up to @xmath102 mt . the strikingly similar results for the different vortex configurations indicate that the @xmath0sr measurements at this small applied field mostly probe the local arrangements of the flux lines , rather than their long - range order .
this is especially seen in the broadening parameter @xmath79 which originally had been introduced in the model to take into account nuclear dipolar broadening and disorder of the vortex lattice .
the temperature dependence of this parameter is almost identical for the different cooling procedures , even though the `` vortex lattice '' should be much more disordered and essentially not clearly defined in the initial @xmath23 measurement , thus strongly suggesting that the broadening of the field distribution is not just related to vortex - lattice disorder .
+ the magnetization measurements do not yield any confirmation for an increased @xmath103 due to the vortex lattice itself on the macroscopic scale the applied field and the magnetization are constant below @xmath104 k ( cf . fig .
[ mvstfc ] ) .
hence , it is unlikely that the sudden increase in @xmath105 below @xmath106 k which is inevitably connected to the rising @xmath103is related to an increase in superfluid density .
moreover , @xmath107 as obtained by the agl model with the freely varying parameters is as low as @xmath108 nm which seems to be far too short for la@xmath1sr@xmath2cuo@xmath3.@xcite furthermore , while the observation of the rather large effective vortex - core radius at @xmath109 is in agreement with previous studies of la@xmath90sr@xmath91cuo@xmath3 , where an expansion of the vortex cores has been found in a relatively low applied field,@xcite the low - temperature values appear to be unreasonably large .
the origin of the increase in the probed mean field , the second central moment of @xmath6 , and the deduced effective vortex - core radius still remains to be clarified .
as shown in fig .
[ zfmusr ] for the presently studied la@xmath1sr@xmath2cuo@xmath3 single crystal zero - field @xmath0sr measurements indicate a slightly enhanced spin depolarization at low temperatures in accordance with the fact that the so - called `` cluster - spin - glass phase '' in la@xmath90sr@xmath91cuo@xmath3 might exist up to a doping level of @xmath110.@xcite therefore , a small increase in the width of the probed field distribution at low temperatures ( @xmath111 k ) could be expected , however , static magnetic phases accounting for the dramatic rise in @xmath92 for @xmath112 k can be excluded from these data .
+ the temperature dependence of @xmath92 is strongly reminiscent of observations by @xmath0sr in bi@xmath9sr@xmath9cacu@xmath9o@xmath10;@xcite there in applied fields of @xmath113 t to @xmath114 t at low temperatures the mean field probed by the muons also grows substantially with decreasing temperature whereas in the higher field of @xmath115 t the probed @xmath92 equals the applied field .
it has been argued that in low fields when the vortex motion `` freezes '' at low temperatures , the random sampling of the spatial field distribution of the vortex lattice is perturbed since pinning centers trap vortices and also offer possible muon stopping sites while at higher fields intervortex interactions dominate.@xcite of course , as already mentioned in sec .
[ sec : magdata ] , it has to be considered that the vortex - lattice topology is different in la@xmath90sr@xmath91cuo@xmath3 and the much more anisotropic bi@xmath9sr@xmath9cacu@xmath9o@xmath10 system .
the observed changes in the mean field in bi@xmath9sr@xmath9cacu@xmath9o@xmath10 fall together with the transition from a `` vortex liquid '' to a `` vortex solid''@xcite whereas the melting of the vortex lattice in the present sample is much closer to @xmath19 as discussed above already .
though , especially in relatively low magnetic fields la@xmath90sr@xmath91cuo@xmath3 shows a rich vortex - matter phase diagram where also in the solid phases the vortices are subject to thermal fluctuations.@xcite therefore , at high temperatures the vortices might `` hop '' thermally activated between different pinning centers which causes the muons only to be exposed to an average field [ thus leading to a broadened but still asymmetric @xmath6 ] while at low temperatures these thermal fluctuations are reduced and more muon spins precess in the high static fields close to the vortex cores which have a substantial effective size in the very low applied magnetic field .
this could explain the observed positive shifts in the first moment and the second central moment of the measured field distribution .
this view is supported by the magnetization data ; even though the relevant time scales differ in the @xmath0sr and magnetization measurements , the observed almost cooling - cycle - independent change in the vortex dynamics around @xmath104 k in the magnetization data as shown in figs .
[ mrelaxation-20mt ] and [ relrates - vs - t ] seems to be correlated with the peculiar variations in the field distributions probed by the muons .
+ sr asymmetry spectra obtained for la@xmath1sr@xmath2cuo@xmath3 at @xmath117 k and @xmath118 k. the low - temperature measurement shows a more exponential - like decay at small times and a slightly increased depolarization rate . ]
very similar to the data discussed here although in higher applied fields
harshman _ et al .
_ observed an inflection point in the temperature dependence of the second central moment of the field distribution measured by tf @xmath0sr in a single crystal of yba@xmath119cu@xmath120o@xmath121.@xcite also there , the increase of @xmath105 at low @xmath122 was attributed to vortex - pinning effects which hence have been taken into account in the analysis .
this analysis , however , does not provide an explanation for the increase in @xmath92 as shown above .
therefore , instead of introducing pinning effects into the model for @xmath83 , in order to accomodate the possible enhanced correlations between the muons and the vortex cores in the analysis , the random sampling of the magnetic fields within the flux - line lattice is replaced by a phenomenological weighting function @xmath123 enhancing the contributions of the fields close to the vortex cores to the resulting @xmath6 . the two - dimensional spatial field distribution calculated using eq .
( [ agl - model ] ) is weighted by the ( not normalized ) lorentzian @xmath124 where @xmath125 represents the distance to the center of the vortex core @xmath126 and @xmath127 is the inverse width of the lorentzian . for the calculation
three to six neighboring vortex cores are taken into account .
the obtained widths of the weighting function are depicted in figs .
[ fch]e , [ scc]e , and [ sch]e , respectively ; above @xmath128 k where @xmath6 is quite narrow , @xmath127 is strongly correlated with @xmath79 but assumes generally small values , meaning that the fields are almost sampled randomly ( @xmath129 ) . for @xmath130
@xmath127 increases only modestly with decreasing temperature until @xmath106 k below which temperature the vortex - core contributions to the overall field distribution grow intensely consistent with the changes in @xmath6 observed during the previous analyses . on the other hand , @xmath92 within
this extended model remains essentially temperature - independent and slightly diamagnetically shifted with respect to the applied field ( cf . figs .
[ fch]b , [ scc]b , and [ sch]b , red circles ) , overall consistent with the magnetization measurements . due to the only small diamagnetic shift of @xmath92 and the scattering of the determined mean - field values
it can not be finally decided if the observed difference in the absolute magnetic moment ( cf .
[ mvstfc ] ) for the employed cooling cycles is reflected in @xmath92 .
( please note that @xmath92 in this case only refers to the mean field connected to the pure randomly sampled ordered vortex lattice , whereas the overall @xmath92 of course is _ not _ independent of temperature as discussed before . ) also , as seen in figs .
[ fch]d , [ scc]d , and [ sch]d the deduced @xmath76 in this analysis is approximately constant as a function of temperature .
the resulting temperature dependence of @xmath82 is shown in figs .
[ fch]a , [ scc]a , and [ sch]a ; it does _ not at all _ show pronounced changes of the curvature around @xmath106 k and is approximately linear in @xmath122 at low temperatures .
the full temperature dependence of @xmath82 can be fairly described by the semi - classical model of chandrasekhar and einzel,@xcite where the presented data are consistent with a weak - coupling @xmath14 order parameter with a maximum zero - temperature gap value @xmath131 and a zero - temperature penetration depth @xmath132 nm ( solid red line in figs .
[ fch]a , [ scc]a , and [ sch]a ) .
the cited uncertainties are of statistical nature only .
especially the value of @xmath107 is subject to sizeable relative systematic errors of at least @xmath133 % due to the phenomenological treatment of the vortex - pinning effects during the analysis .
nevertheless , the so determined value of @xmath134 agrees well with data found in the literature @xcite while it appears to be estimated much too short if a randomly sampled vortex lattice was assumed as discussed at the end of sec .
[ sec : musrdata ] .
+ it should be noted that the presented explanation for the observed field distributions might not be the only possible one yet , the introduced phenomenological model describes all obtained data consistently even though up to now the weight @xmath127 can not be fully related to the underlying quantities such as the pinning potential .
any other model would also have to be coherent and particularly it would have to provide an explanation for the temperature dependence of @xmath92 .
+ one such other possibility to explain the @xmath0sr data would be the presence of field- or vortex - induced antiferromagnetic order which would not be visible in the macroscopic magnetization measurements .
this phenomenon has been reported to be present in cuprate high - temperature superconductors in applied fields of a few tesla by various techniques.@xcite yet , based on neutron - scattering results which indicate no ordered magnetic phase in low fields in la@xmath90sr@xmath91cuo@xmath3 with such high doping level as used in the present study , sonier _ et al . _
concluded from their @xmath0sr studies that only disordered field - induced magnetism is likely to occur in low applied fields.@xcite therefore , the presence of an ordered magnetic state can be excluded for the presently discussed data .
the tf @xmath0sr data presented in this work show a striking resemblance with data obtained earlier by khasanov _ et al . _ in @xmath0sr experiments on single crystals of cuprate high - temperature superconductors .
their data have been interpreted as evidence for two distinct energy gaps in the quasiparticle excitation spectrum , e.g. of la@xmath1sr@xmath2cuo@xmath3,@xcite yba@xmath119cu@xmath120o@xmath121,@xcite and yba@xmath9cu@xmath3o@xmath135.@xcite note that this is basically different from reports of slight deviations from a @xmath14 gap symmetry as measured , e.g. by phase - sensitive methods in yba@xmath119cu@xmath120o@xmath121.@xcite the conclusions of ref
. are mainly based on the observation of an inflection point in the temperature dependence of the second central moment of the magnetic field distribution @xmath6 obtained by tf @xmath0sr measurements in the vortex state ( like seen in fig .
[ fch]a ) , while the corresponding temperature dependence of the measured mean field has not been discussed at all .
our tf @xmath0sr data on la@xmath1sr@xmath2cuo@xmath3 are very similar , yet , the above presented analysis within the framework of conventional vortex - lattice - generated field distributions extended by vortex - pinning effects appears to describe the overall @xmath0sr and magnetization data in a coherent way .
this is also in line with the suppression of the low - temperature increase of the second central moment of @xmath6 in higher applied fields@xcite where inter - vortex interactions compete with vortex pinning and minimize its effects on the measured field distributions .
+ and , while it seems to be a reasonable approach to model two distinct gaps if an inflection point in @xmath136 is found ( e.g. @xmath137@xcite or @xmath138@xcite ) , the observation of the strong increase in the probed mean field @xmath103 in the tf @xmath0sr measurements which is not observed in the magnetization measurements , renders it unlikely that the observed full second central moment of @xmath6 is a true measure of @xmath139 at low temperatures .
moreover , it is pointed out , that the reported agreement between the obtained magnitudes of two - gap contributions from tf @xmath0sr and neutron crystal - field spectroscopy@xcite is probably fortuitous since fundamentally different models were employed to obtain those values .
while in the analysis of the tf @xmath0sr data the contributions to the superfluid density were treated to be additive ( _ after _ the integrations over the fermi surface),@xcite in the analysis of the neutron data the individual gaps are added together ( _ before _ the integration over the fermi surface ) yielding effectively _
one pseudogap _ with an anisotropy @xmath140 and shifted nodes.@xcite taking into account such an anisotropic _ superconducting gap _ in the modeling of the superfluid density would result only in tiny overall changes from the @xmath14 scenario since firstly , only the modulus of the gap is relevant and secondly , the integration over the full fermi surface@xcite averages out small effects .
therefore , in general , a small gap anisotropy in the sense of refs . or can not be excluded , however , its determination from magnetic - penetration - depth data is rather difficult and indirect .
+ in conclusion , using low - field tf @xmath0sr in the mixed state of a la@xmath1sr@xmath2cuo@xmath3 single crystal a strong increase in the probed mean field and the second central moment of the measured field distribution is found at low temperatures . by combining @xmath0sr and magnetization measurements it is shown that these effects seem to be primarily related to vortex - pinning effects changing the sampling of the spatial field distribution by the muons and that the data can be consistently described by taking into account a single energy gap in the quasiparticle excitation spectrum with @xmath14 symmetry . however
, small variations of the superconducting order parameter can not be excluded on the basis of the presented @xmath0sr data . given this assessment
, the extrapolated magnetic penetration depth @xmath141 nm agrees well with reports of earlier experiments .
+ furthermore , this article commemorates that tf @xmath0sr in the way it has been employed for the present study is first and foremost a very sensitive technique for the determination of local static or dynamic magnetic field distributions but the muon is neither a _ direct _ probe of the superfluid density of a superconductor in the mixed state nor of its order - parameter symmetry as also has been realized earlier.@xcite nevertheless , these parameters of a type - ii superconductor can be deduced from tf @xmath0sr data in the vortex state if a coherent description of the determined field distributions is available .
the @xmath0sr measurements were performed at the swiss muon source , paul scherrer institute , villigen , switzerland .
the @xmath0sr time spectra have been analyzed using the free software package ` musrfit`@xcite mainly developed by a. suter and based on the cern ` root ` framework@xcite including the ` minuit ` routines for function minimization.@xcite we thank m. bendele for his support during the @xmath0sr experiments .
helpful discussions with a. suter , a. furrer , e. h. brandt , and m. bendele are gratefully acknowledged .
this work has been supported by the swiss national science foundation and the nccr manep .
s. l. lee , p. zimmermann , h. keller , m. warden , i. m. savi , r. schauwecker , d. zech , r. cubitt , e. m. forgan , p. h. kes , t. w. li , a. a. menovsky , and z. tarnawski , phys .
* 71 * , 3862 ( 1993 ) .
m. weber , a. amato , f. n. gygax , a. schenck , h. maletta , v. n. duginov , v. g. grebinnik , a. b. lazarev , v. g. olshevsky , v. yu . pomjakushin , s. n. shilov , v. a. zhukov , b. f. kirillov , a. v. pirogov , a. n. ponomarev , v. g. storchak , s. kapusta , and j. bock , phys .
b * 48 * , 13022 ( 1993 ) .
r. kadono , w. higemoto , a. koda , m. i. larkin , g. m. luke , a. t. savici , y. j. uemura , k. m. kojima , t. okamoto , t. kakeshita , s. uchida , t. ito , k. oka , m. takigawa , m. ichioka , and k. machida , phys .
b * 69 * , 104523 ( 2004 ) .
u. divakar , a. j. drew , s. l. lee , r. gilardi , j. mesot , f. y. ogrin , d. charalambous , e. m. forgan , g. i. menon , n. momono , m. oda , c. d. dewhurst , and c. baines , phys .
* 92 * , 237004 ( 2004 ) .
j. e. sonier , f. d. callaghan , y. ando , r. f. kiefl , j. h. brewer , c. v. kaiser , v. pacradouni , s. a. sabok - sayr , x. f. sun , s. komiya , w. n. hardy , d. a. bonn , and r. liang , phys .
b * 76 * , 064522 ( 2007 ) . | muon - spin rotation ( @xmath0sr ) experiments are often used to study the magnetic field distribution in type - ii superconductors in the vortex state . based on the determination of the magnetic penetration depth
it is frequently speculated also controversially about the order - parameter symmetry of the studied superconductors .
this article reports on a combined @xmath0sr and magnetization study of the mixed state in the cuprate high - temperature superconductor la@xmath1sr@xmath2cuo@xmath3 in a low magnetic field of @xmath4 mt applied along the @xmath5 axis of a single crystal .
the macroscopic magnetization measurements reveal substantial differences for various cooling procedures . yet , indicated changes in the vortex dynamics between different temperature regions as well as the results of the microscopic @xmath0sr experiments are virtually independent of the employed cooling cycles . additionally , it is found that the mean magnetic flux density , locally probed by the muons , strongly increases at low temperatures .
this can possibly be explained by a non - random sampling of the spatial field distribution of the vortex lattice in this cuprate superconductor caused by intensified vortex pinning . |
from the conceptual point of view different degrees of synchronization can be distinguished : complete synchronization @xcite , generalized synchronization @xcite , lag synchronization @xcite , phase synchronization @xcite , and burst ( or train ) synchronization @xcite . in the following ,
we focus our attention on phase synchronization in stochastic systems that has attracted recent interest for the following reason : in many practical applications the dynamics of a system , though not perfectly periodic , can still be understood as the manifestation of a stochastically modulated limit cycle @xcite . as examples , we mention neuronal activity @xcite , the cardiorespiratory system @xcite , or population dynamics @xcite .
given a data set or some model dynamics there exists a variety of methods to define an instantaneous phase @xmath3 of a signal or a dynamics . for a clear cut separation of deterministic and noise - induced effects it is essential to assess the robustness of each of these different phase definitions with respect to noise .
section [ sec:1 ] is devoted to this issue .
since we do not distinguish between dynamical and measurement noise @xcite our treatment is also tied to the question how synchronization can be detected within any realistic experimental data .
the synchronization properties of a noisy system can be classified in a hierarchical manner : stochastic phase locking always implies frequency locking while the converse is not true in general . on the other hand ,
small phase diffusivity is necessary but not sufficient for phase synchronization .
this will become clear in sec .
[ sec:2 ] when we review an analytic approach @xcite to stochastic phase synchronization developed for a thermal two - state system with transitions described by noise - controlled rates .
a recently proposed method @xcite to measure the average phase velocity or frequency in stochastic oscillatory systems based on rice s rate formula for threshold crossings @xcite will be presented and discussed in sec .
[ sec:3 ] .
the rice frequency proves to be useful especially in underdamped situations whereas the overdamped limit yields only finite values for coloured noise .
its relation to the frequency based on the widely used hilbert phase ( cf .
[ sec:1:phih ] ) is discussed and illustrated . in the final sec .
[ sec:4 ] we connect the topic of stochastic resonance ( sr ) @xcite with results on noise - enhanced phase coherence @xcite . to this end
we study the synchronization properties of the bistable kramers oscillator driven externally by a periodic signal @xcite . as a complement to the frequently investigated overdamped limit
, we consider here the underdamped case employing the methods presented in sec .
[ sec:3 ] .
a phase occurs in a quite natural way when describing the cyclic motion of an oscillator in phase space .
self - sustained oscillators are nonlinear systems that asymptotically move on a limit cycle .
the instantaneous position in phase space can be represented through instantaneous amplitude @xmath5 and phase @xmath6 .
a systematic approach to relate the amplitude and phase dynamics to the dynamics formulated in original phase space was developed by bogoliubov and mitropolski @xcite .
their method starts from the following decomposition of the dynamics @xmath7 where the function @xmath8 comprises all terms of higher than first order in @xmath9 ( nonlinearities ) , velocity dependent terms ( friction ) , and noise . in their work bogoliubov and mitropolski
considered the function @xmath8 to be a small perturbation of order @xmath10 which means that the system is weakly nonlinear and the noise or the external forces are comparatively small as not to distort the harmonic signal too much .
the definition of an instantaneous phase proceeds by expressing the position @xmath9 and the velocity @xmath11 in polar coordinates @xmath12 and @xmath4 @xmath13\,,\\ \label{eq : bogmist2b } v(t ) & = & -\omega_0\ ; a^n(t ) \sin\left[\phi^n(t)\right]\end{aligned}\ ] ] which yields by inversion @xcite @xmath14 ^ 2 } \,,\\ \label{eq : bogmist3b } \phi^n(t ) & = & \arctan \left [ -\frac{v(t)/{\omega}_0}{x(t ) } \right ] .\end{aligned}\ ] ]
it should be noted that a meaningful clockwise rotation in the @xmath15-plane determines angles to be measured in a specific way depending on the sign of @xmath16 .
using eqs .
( [ eq : bogmist2a ] ) , ( [ eq : bogmist2b ] ) , ( [ eq : bogmist3a ] ) and ( [ eq : bogmist3b ] ) it is straightforward to transform the dynamics in @xmath9 and @xmath11 , eqs .
( [ eq : bogmist1a ] ) and ( [ eq : bogmist1b ] ) , into the following dynamics for @xmath12 and @xmath4 @xcite @xmath17 the line @xmath18 corresponds to angles @xmath19 . as can be read off from eq .
( [ eq : phidot ] ) , the phase velocity always assumes a specific value for @xmath18 @xcite , i.e. , @xmath20 this has the following remarkable consequence .
we see that even in the presence of noise passages through zero in the upper half plane @xmath21 are only possible from @xmath22 to @xmath23 , in the lower half plane only from @xmath23 to @xmath22 .
this insight becomes even more obvious from a geometrical interpretation : as the noise exclusively acts on the velocity @xmath11 , cf .
( [ eq : bogmist1b ] ) , it can only effect changes in the vertical direction ( in @xmath15-space ) . along the vertical line @xmath18
, however , the angular motion possesses no vertical component while radial motion is solely in the vertical direction and , therefore , only affected by the noise . from this
we conclude that between subsequent zero crossings of the coordinate with positive velocity the phase has increased by an amount of @xmath24 .
this finding establishes a simple operational instruction how to measure the average phase velocity of stochastic systems .
we will come back to this point in sec .
[ sec:3 ] as we have just seen zero crossings can be utilized to mark the completion of a cycle .
this can be generalized to the crossings of an arbitrary threshold with positive velocity or even to the crossing of some separatrix . in this connection
the concept of isochrones of a limit cycle has to be mentioned @xcite .
all of these extensions of the natural phase require a thorough knowledge of the dynamics and the phase space structure .
in many practical applications , however , the detailed phase portrait is not known . instead
, one is given a data series exhibiting a repetition of characteristic marker events , e.g. the spiky peaks of neural activity , the r - peaks of an electrocardiogram , or pronounced maxima as found in population dynamics .
these marker events can be used to pinpoint the completion of a cycle , @xmath26 , and the beginning of a subsequent one , @xmath27 .
it is then possible to define an instantaneous phase @xmath28 by linear interpolation , i.e. , @xmath29 where the times @xmath30 are fixed by the marker events . reexpressing the time series @xmath31 of the system as @xmath32,\ ] ]
then defines an instantaneous amplitude @xmath33 .
the benefit of such a treatment is to reveal a synchronization of two or more such signals : whereas the instantaneous amplitudes and , therefore , the time series might look rather different , the phase evolution can display quite some similarity .
if the average growth rates of phases match ( notwithstanding the fact that phases may diffuse rapidly ) the result is termed frequency locking .
small phase diffusion , in addition to frequency locking , means that phases are practically locked during long episodes that occasionally are disrupted by phase slips caused by sufficiently large fluctuations .
this elucidates the meaning of effective phase synchronization in stochastic systems .
as should be clear from its definition the linear phase relies on the clear identification of marker events . with increasing noise intensity
this identification will fail since sufficiently large fluctuations may either mask true or imitate spurious marker events . on the other hand , in some cases , e.g. for excitable systems
, fluctuations can be essential for the generation of marker events , i.e. , marker events may be noise - induced .
as a final remark , let us mention that relative maxima of a differentiable signal @xmath31 correspond to positive - going zeros of its derivative @xmath34 .
however , this seemingly trivial connection is overshadowed by complications if the derivative itself is a non - smooth function that does not allow to easily extract the number of zero crossings ( cf .
[ sec:3 ] and fig .
[ f_oscillator ] ) . in situations
where a measured signal @xmath31 exhibits a lot of irregularity it is not quite clear how to define a phase the signal might look far from a perturbed harmonic or even periodic one and marker events can not be identified unambiguously .
the concept of the analytic signal as introduced by gabor @xcite offers a way to relate the signal @xmath31 to an instantaneous amplitude @xmath36 and a phase @xmath37 .
the physical relevance of a such constructed phase is a question of its own ; for narrow - band signals or harmonic noise it has a clear physical meaning whereas the general case requires further considerations ( cf .
appendix a2 in @xcite ) .
the analytic signal approach extends the real signal @xmath31 to a complex one @xmath38 $ ] with the imaginary part @xmath39 resulting from an appropriate transformation of the real part @xmath31 . instead of taking @xmath40 as for the natural phase we search @xmath39 as the result of a convolution of @xmath31 with some appropriate kernel @xmath41 , i.e. , @xmath42 now , _
means that the kernel has to be chosen such that the method reproduces the phase of a harmonic signal .
applying the convolution theorem it is easy to see that the fourier transform @xmath43 of the transformed signal @xmath39 should be related to the fourier transform @xmath44 of the original signal @xmath31 by a phase shift that transforms a cosine into a sine , i.e. , @xmath45 where sgn(@xmath46 ) is the sign function . by an inverse fourier
transform we thus find that @xmath47 which implies that @xmath39 is related to @xmath31 via the hilbert transform @xmath48 the symbol @xmath49 in front of the integral in eq .
( [ eq : hilberttrafo ] ) is a reminder that the integral has to be evaluated in the sense of the cauchy principal value .
the fact that the hilbert phase @xcite @xmath50\ ] ] arises as the result of a convolution instead of a differentiation makes it less sensitive to short - lived small fluctuations .
this observation was already reported by vainstein and vakman @xcite .
moreover , the construction by a convolution brings the hilbert phase in close contact with the wavelet transform that is widely used to compute a time - dependent spectral decomposition of non - stationary signals @xcite . by virtue of eq .
( [ eq : fourhilb ] ) it is evident that the fourier transform of the analytic signal @xmath51 satisfies the relation @xmath52\,x(\omega ) = 2\,\theta(\omega)\,x(\omega)\ ] ] where @xmath53 is the step function .
equation ( [ eq : anasigfourier ] ) shows that all positive fourier components of the original signal contribute with equal weights .
selecting , instead , a certain frequency band by using a subsequent gaussian filter ( in frequency space ) corresponds to a convolution with the morlet wavelet ( gabor function ) .
generalized phase definitions and the wavelet transform were employed to detect phase - synchronous activity in the brain @xcite and in a chaotic laser array @xcite .
multistability is one of the crucial consequences of nonlinearity and plays a dominant role for many important topics , e.g. evolution , information processing and communication , pattern formation , etc .
frequently , fluctuations play a benificial role in that they effect transitions between the different states that are directly tied to the performance of a task . the phenomenon of sr @xcite , for instance , can be observed in a bistable system .
typically , the information that is processed in a bistable system does not require to keep track of a continuum but is rather contained in the switching events between the two states @xcite .
hence , it is desirable to link the dichotomous switching process to a description in terms of an instantaneous phase . since switching to and fro constitutes one cycle and , thus , corresponds to a phase increment of @xmath24 the linear interpolating phase @xmath25 can be readily constructed employing switchings as marker events of half - cycles .
furthermore , the hilbert phase can be easily computed .
alternatively , it is possible to use the switching process ( between @xmath55 and @xmath56 ) and construct a discrete instantaneous phase @xmath57 changing discontinuously at the switching events .
the last - mentioned instantaneous phase is obtained simply by multiplying the number of switches ( or the number of renewals in renewal theory @xcite ) @xmath58 by the value @xmath59 , i.e. , @xmath60 .
the instantaneous state , in turn , can be obtained from the discrete phase via @xmath61 $ ] . in fig .
[ fig:3phases ] we show how the three alternative phases @xmath62 and @xmath54 agree in the description of a dichotomous switching process . note that the natural phase @xmath6 is related to the underlying process in real phase space and , hence , can not be deduced from the two - state signal .
the advantage of the discrete phase is that it allows an analytic treatment of effective phase synchronization in stochastic bistable systems .
this will be addressed in sec .
[ sec:2 ] .
the robustness of the discrete phase with respect to noise comes into play not when making the `` transition '' from the switching process to the instantaneous phase @xmath57 but when constructing the switching events from the continuous stochastic trajectory . at the level of a markovian switching process noise
enters only via its intensity that changes transition rates . in systems of dimension @xmath63
there are various ways how to define one or even more instantaneous phases : projecting the @xmath64-dimensional phase space onto 2-dimensional surfaces , choosing poincar sections or computing the hilbert phase for each of the coordinates .
many of the methods can only be done numerically and always require to consider the dynamics in detail in order to check whether a made choice is appropriate .
we will not elaborate these details here but refer to @xcite ( especially chap .
10 ) and references therein .
in this section we consider a stochastic bistable system , for instance a noisy schmitt trigger , which is driven externally either by a dichotomous periodic process ( dpp ) or a dichotomous markovian process ( dmp ) .
the dichotomous character of the input shall be either due to a two - state filtering or be rooted in the generation mechanism of the signal .
the bistable system generates a dichotomous output signal . for convenience
we choose to label input and output states with values @xmath56 and @xmath55 respectively .
the dpp is completely specified by its angular frequency @xmath65 where @xmath2 is the half - period .
accordingly , the dmp is fully characterized by its average switching rate @xmath66 . since both input and output are two - state variables it is possible to study phase synchronization in terms of discrete indicating the discrete phase considered throughout this section ] input and output phases @xmath67 and @xmath68 respectively .
consequently , also the phase difference @xmath69 is a discrete quantity that can assume positive and negative multiples of @xmath59 . from the definition of the phase difference
@xmath70 it follows that each transition between the output states _ increases _
@xmath70 by @xmath59 whereas each transition between the input states _ reduces
_ @xmath70 by @xmath59 .
transitions between the input states are governed by the rates @xmath71 where a single realization of the dpp is characterized by deterministic switching times @xmath72 . here
, @xmath73 is the initial phase of the input signal rendering the periodic process non - stationary ( cyclo - stationary ) . to achieve strict stationarity we average the periodic dynamics with respect to @xmath73 which is equidistributed over the interval @xmath74 . in the absence of an input signal
the two states are supposed to be symmetric and the hopping rates for both directions are identical and completely determined by a prefactor @xmath75 , the energy barrier @xmath76 , and the noise intensity @xmath77 .
the central assumption of our analysis is that the input signal modifies the transition rates of the output solely through the phase difference @xmath70 in the following way @xmath78\,,\ ] ] where the function @xmath79 and the amplitude @xmath80 to keep the signal subthreshold .
this definition introduces two noise - dependent time scales @xmath81 with @xmath82 . the function @xmath83 favours phase differences with even multiples of @xmath59 , i.e. , in - phase configurations .
a description of the stochastic evolution of the phase difference is based on the probabilities @xmath84 to experience a phase difference @xmath70 at time @xmath85 conditioned by a phase difference @xmath86 at time @xmath87 . due to the discrete character of @xmath70 ( allowing only for multiples of @xmath59 ) we briefly denote @xmath88
. then the probabilistic evolution operator reads with @xmath89 from ( [ kram1 ] ) @xmath90 while the last two terms on the right hand side of eq .
( [ ph1 ] ) account for the change of @xmath70 due to transitions of the output the operator @xmath91 reflects switches of the input @xmath92 with the related input switching rates @xmath93 given by eq .
( [ inputrates ] ) . as mentioned above the non - stationary ( cyclo - stationary ) character of the dpp can be cured by averaging over the initial phase @xmath73 . since `` temporal '' and `` spatial '' contributions in eq .
( [ l ] ) are separable we can perform this average prior to the calculation of any moment of @xmath70 yielding @xmath94 from eq .
( [ average ] ) we see that the @xmath73-averaged dpp formally looks equivalent to a dmp with the transition rate @xmath95 .
of course , initial phase averaging does not really turn a dpp into a dmp .
the subtle difference is that while @xmath70-moments of the dmp continuously change in time related @xmath70-averages of the dpp ( before the @xmath73-average ) are still discontinuous , hence , temporal derivatives of functions of @xmath96 have to be computed with care before initial phase averaging @xcite . using standard techniques @xcite we can derive the evolution equation for the mean phase difference @xmath96 from eq .
( [ ph1 ] ) @xmath97 here , @xmath98 denotes the average frequency of the input phase and equals @xmath99 for the dmp and @xmath100 for the dpp . assuming higher moments uncoupled , i.e. , @xmath101 , eq .
( [ ph2b ] ) is adler s equation @xcite arising in the context of phase locking . note that here both the frequency mismatch @xmath102 and the synchronization bandwidth @xmath103 are noise dependent .
this elucidates the opportunity to achieve _ noise - induced _ frequency and effective phase locking . for the short - time evolution
a necessary condition for locking is @xmath104 which defines `` arnold tongues '' @xcite of synchronization in the @xmath105 vs. @xmath77 plane .
the kinetic equation for @xmath106 can be evaluated explicitly yielding @xmath107 \mean{\sigma } + a_2-a_1\,.\ ] ] from eq .
( [ sigdot ] ) we see that @xmath106 approaches a stationary value @xmath108 that exactly coincides with the stationary correlation coefficient between the input and output @xcite .
the relaxation time is given by @xmath109^{-1}$ ] .
hence , the stationary output phase velocity can be achieved from eq .
( [ ph2b ] ) by insertion of eq .
( [ ph4 ] ) yielding @xmath110 this expression is in agreement with similar results derived in the context of resonant activation @xcite . in fig .
[ fig : epl1 ] we depict the mean output switching rate @xmath111 as a function of the noise intensity @xmath77 for several values of the input signal amplitude @xmath105 . with increasing amplitude the region of frequency locking , i.e. , the region of @xmath77 for which @xmath112 , widens ( cf .
also fig . 9 in @xcite ) . for most of these intensities
the bistable system possesses rates that do not obey the time scale matching condition ; nevertheless , on the average the output switching events are entrained by the input signal . as mentioned
before the whole effect is non - linear : if detuning becomes too large the output gets desynchronized and returns to its own dynamics . besides many alternatives , we defined the width of the frequency locking region by requiring the slope of the curves to be less than 30% of the slope for @xmath113 at the point where the output switching rate coincides with @xmath114 [ or @xmath95 respectively ] simultaneously disqualifying the initial flat region for small @xmath77 .
the resulting `` arnold tongues '' ( compatible with data from @xcite ) are shown in the inset of fig .
[ fig : epl1 ] .
it can be seen that frequency locking necessitates to exceed a minimal amplitude that shifts to lower values for slower signals @xcite .
let us emphasize that the frequency locking region groups around the noise intensity that satisfies the time scale matching condition ( in our case by definition @xmath115 ) ; for a harmonic input this range of @xmath77 also maximizes the spectral power amplification , in contrast to values of @xmath77 where the signal - to - noise ratio attains its maximum ( @xmath116 ) @xcite .
the phenomenon of phase locking can be demonstrated by considering the diffusion coefficient @xmath117 of the phase difference , achieved as the time derivative of the variance @xmath118 $ ] . performing the calculation for both the dmp and dpp yields
@xmath119 \label{d*p}\ ] ] with @xmath120 for the dmp and @xmath121 for the dpp .
the stationary correlator @xmath122 , i.e. , the asymptotic limit of @xmath123 , can be computed from the corresponding kinetic equation @xcite .
inserting @xmath122 into eq .
( [ d*p ] ) we thus find for the dmp ( cf .
[ fig : epl2 ] top ) @xmath124 \label{dmp2}\end{aligned}\ ] ] and for the dpp ( cf . fig .
[ fig : epl2 ] bottom ) @xmath125 \label{dpp2}\end{aligned}\ ] ] with @xmath126 given by eq .
( [ ph4 ] ) and @xmath111 by eq .
( [ ph5 ] ) . both eqs .
( [ dmp2 ] ) and ( [ dpp2 ] ) possess the same structure @xmath127 with @xmath128 .
since @xmath129 is never decreasing ( with increasing noise intensity ) the same is true for the sum of the first two terms .
the possibility of synchronized input - output jumps is rooted in @xmath130 . since
this term comprises only contributions scaling with powers of @xmath131 , which itself rapidly vanishes for small @xmath77 ( cf .
inset of fig .
[ fig : epl2 ] bottom ) , we first observe an increase of @xmath132 .
an increase of @xmath131 signals the coherent behaviour of input and output and , consequently , endows @xmath130 with considerable weight to outbalance the increase of @xmath129 . as can be seen from the inset of fig .
[ fig : epl2 ] bottom a negative slope is initiated only for a sufficiently large @xmath131 .
however , the range of high input - output correlation @xmath131 does not determine the range of low diffusion coefficients since at rather high noise intensities @xmath77 the output switches with a large variance and thus , finally dominates over the ordering effect of @xmath133 .
plotting the boundaries of the region where @xmath134 defines the tongues depicted in the inset of fig .
[ fig : epl2 ] top . as for the `` arnold tongues '' in fig .
[ fig : epl1 ] a minimal amplitude varies with the mean input switching rate @xmath98 and shifts to lower values when considering slower signals .
it is worth mentioning that the addition of an independent dichotomous noise that modulates the barrier @xmath76 can drastically reduce this minimal amplitude if this second dichotomous noise switches faster than the external signal @xcite .
the minimum of @xmath117 observed in the region of frequency locking can be equivalently expressed as a pronounced maximum of the average duration of locking episodes @xmath0 . to show this we note that a locking episode is ended by a phase slip whenever the phase difference has changed , i.e. , increased or decreased , by the order of @xmath59 , or @xmath135 this quadratic equation
can be solved for @xmath0 @xcite and by inserting the noise - dependent expressions for @xmath136 and @xmath117 we can compute @xmath137 as a function of noise intensity @xmath77 where @xmath2 is either @xmath138 for the dmp or @xmath139 for the dpp respectively .
the results for both the dmp and the dpp are plotted in fig .
[ fig : tlock ] .
a pronounced maximum for intermediate values of noise intensity clearly proves that noise - induced frequency synchronization is accompanied by noise - induced phase synchronization .
as mentioned in sec . [ sec:1:phin ] positive - going zero crossings can be used to count completions of a cycle in oscillatory systems . in this view the average frequency ,
i.e. , the average phase velocity , turns out to be the average rate of zero crossings which is captured by a formula put forward by rice @xcite .
this elementary observation yields a novel way to quantify the average frequency of a phase evolution , henceforth termed the `` rice frequency '' , and to prove frequency locking in stochastic systems . to detail our derivation of the rice frequency in this section , we start from the following one - dimensional potential system @xmath140 subjected to gaussian white noise @xmath141 of intensity @xmath77 , i.e. , @xmath142 and being driven by the external harmonic force @xmath143 . in fig .
[ f_oscillator ] we show a sample path for the harmonic oscillator @xmath144 where we used the friction coefficient @xmath145 , the natural frequency @xmath146 , and a vanishing amplitude @xmath147 of the external drive . as can be read off from fig .
[ f_oscillator ] , the velocity @xmath148 basically undergoes a brownian motion and , therefore , constitutes a rather jerky continuous , but generally not differentiable signal . in particular ,
near a zero crossing of @xmath11 there are many other zero crossings .
in contrast to that , the coordinate @xmath9 is a much smoother signal since it is determined by an integral over a continuous function @xmath149 and , therefore , differentiable . in particular , near a zero crossing of @xmath9 there are no other zero crossings . in the following , we will take advantage of this remarkable smoothness property of @xmath9 that is an intrinsic property of the full oscillatory system ( [ e_osc ] ) and disappears when we perform the overdamped limit . in 1944
, rice @xcite deduced a formula for the average number of zero crossings of a smooth signal like @xmath9 in the oscillator equation ( [ e_osc ] ) . in this rate formula
enters the probability density @xmath150 of @xmath9 and its time derivative , @xmath148 , at a given instant @xmath85 . the rice rate for passages through zero with positive slope ( velocity )
is determined by @xcite @xmath151 this time - dependent rate is to be understood as an ensemble average . if the dynamical system is ergodic and mixing the asymptotic stationary rate @xmath152 can likewise be achieved by the temporal average of a single realization .
let @xmath153)$ ] be the number of positive - going zeros of the signal @xmath9 in the time interval @xmath154 $ ] .
using ergodicity , the relation @xmath155\right ) } { t}\ ] ] is fulfilled for the process characterized by the stationary density @xmath156 . in the following we always consider stationary quantities . as explained in sec .
[ sec:1:phin ] , the zero crossings can be used as marker events to define an instantaneous phase @xmath28 by linear interpolation , cf .
( [ linphase ] ) .
the related average phase velocity is the product of the ( stationary ) rice rate and @xmath24 and , hence , called the ( stationary ) rice frequency @xmath157 for a dynamics described by a potential @xmath158 in the absence of an external driving , i.e. , ( [ e_osc ] ) with @xmath147 , the stationary density can be calculated explicitly yielding @xmath159\ ] ] where @xmath160 is the normalization constant . from this and the application of eq .
( [ eq : ricefreq ] ) , it is straightforward to derive the exact result @xmath161 } { \int\limits_{-\infty}^{\infty}\exp\left[-{u(x)\over d}\right]\,dx}\,.\ ] ] without loss of generality we can set @xmath162 . in the limit @xmath163 , we can perform a saddlepoint approximation around the deepest minima @xmath164 ( e.g. for symmetric potentials ) . in this way we find the following expression valid for @xmath165 , i.e. , the small noise approximation , @xmath166\over\sqrt{u^{''}(x_i ) } } \right]^{-1}\,.\ ] ] in the limit @xmath167 , we have to consider the asymptotic behaviour of the potential , @xmath168 , to estimate the integral in eq .
( [ eq : potsys ] ) . for potentials that can be expanded in a taylor series about zero and that , therefore , result in a power series of order @xmath169 , i.e. , @xmath170 , we can rescale the integration variable by @xmath171 . for sufficiently large @xmath77 , the integral
is dominated by the power @xmath169 term . in this way
we find the large noise scaling @xmath172 applying eqs .
( [ eq : potsys ] ) and ( [ eq : smallnoise ] ) to the harmonic oscillator ( [ e_harmonicosc ] ) we immediately find that @xmath173 , independent of @xmath114 and for all values of @xmath174 .
this is also in agreement with eq .
( [ eq : largenoise ] ) .
it follows because @xmath175 implies that , for large noise , the rice frequency @xmath176 does not depend on @xmath77 at all .
note , however , that in the deterministic limit , i.e. , for @xmath177 , we have the standard result @xmath178 which explicitly does depend on the friction strength @xmath179 .
therefore , the limit @xmath177 is discontinuous except in the undamped situation @xmath180 .
the similarity of eqs .
( [ eq : potsys ] ) and ( [ eq : smallnoise ] ) with rates from transition state theory @xcite will be addressed below when we discuss the bistable potential .
it is well known that the rice frequency can not be defined for stochastic variables that integrate increments of the wiener process ( white noise ) . from eq .
( [ e_osc ] ) this holds true for the velocity @xmath181 .
this is so , because the stochastic trajectories of degrees of freedom being subjected to gaussian white noise forces are continuous but are of _ unbounded _ variation and nowhere differentiable @xcite .
this fact implies that such stochastic realizations cross a given threshold within a fixed time interval infinitely often if only the numerical resolution is increased _
ad infinitum_. this drawback , which is rooted in the mathematical peculiarities of idealized gaussian white noise , can be overcome if we consider instead a noise source possessing a finite correlation time , i.e. , coloured noise , see ref .
@xcite . to this end
, we consider here an oscillatory noisy harmonic dynamics driven by gaussian exponentially correlated noise @xmath182 , i.e. , @xmath183 with @xmath182 obeying @xmath184 and @xmath185 following the same reasoning as before we find for the rice frequency of @xmath31 as before @xmath186 likewise , upon noting that within a time interval @xmath187 , or @xmath188 , respectively , the rice frequency of the zero crossings with positive slope of the process @xmath189 is given by @xmath190 which is evaluated to read @xmath191 the result in ( [ eq : wx ] ) shows that for small noise colour @xmath192 the rice frequency for @xmath193 assumes a correction @xmath194 , as @xmath195 . in clear contrast
, the finite rice frequency for the velocity process @xmath189 ( [ eq : col1b ] ) diverges in the limit of vanishing noise colour proportional to @xmath196 . to exemplify the relation between the rice frequency @xmath176 and the hilbert frequency @xmath197 ,
again we consider the damped harmonic oscillator eq .
( [ e_harmonicosc ] ) agitated by noise alone . in fig .
[ f_hilbertoscillator ] we show a numerically evaluated sample path and the corresponding hilbert phase ( normalized to @xmath24 and modulo @xmath198 ) using the parameters @xmath199 .
an important point to observe here is that around @xmath200 and @xmath201 the hilbert phase @xmath35 does not increase by @xmath24 after two successive passages through zero with positive slope .
this shall illustrate the difference between the hilbert phase and the natural phase .
in subsec .
[ sec:1:phih ] this observation was already mentioned as a consequence of the nonlocal character of the hilbert transform . in particular , short and very small
amplitude crossings to positive @xmath9 are not properly taken into account by the hilbert phase since they only result in a small reduction of @xmath35 .
this leads us to conjecture that quite generally @xmath202 holds .
in fact , for the case of the harmonic oscillator that generates a stationary gaussian process one even can prove this conjecture by deriving explicit expressions for @xmath176 and @xmath197 . as usual , let @xmath203 denote the spectrum of the stationary gaussian process @xmath9 .
then the rice frequency can be recast in the form of @xcite @xmath204^{1/2}\ , .
\label{e_srice}\ ] ] a similar expression ( additionally involving an arrhenius - like exponential ) exists when considering not zero crossings , as in eq .
( [ e_srice ] ) , but crossings of an arbitrary threshold . in @xcite
it was shown that the hilbert frequency of the same process @xmath9 is given by a similar expression , namely @xmath205\ , .
\label{e_shil}\ ] ] interpreting the quantity @xmath206 as a probability density @xmath207 , @xmath208 , we can use the property that the related variance is positive , i.e. , @xmath209 ^2 \,.\ ] ] taking the square - root on both sides of the last inequality immediately proves eq .
( [ e_ungleich ] ) . using the spectrum of the undriven noisy harmonic oscillator @xmath210 and employing eqs .
( [ e_srice ] ) and ( [ e_shil ] ) , it is easy to see that both @xmath176 and @xmath197 do not vary with @xmath77 .
we have already shown above that @xmath211 .
in contrast to this , @xmath197 is a monotonically decreasing function of @xmath114 that approaches @xmath212 from below in the limit @xmath213 .
the probability density of the periodically driven noisy harmonic oscillator can be determined analytically by taking advantage of the linearity of the problem .
introducing the mean values of the coordinate and the velocity , @xmath214 and @xmath215 , the variables @xmath216 obey the differential equation of the undriven noisy harmonic oscillator .
in the asymptotic limit @xmath217 the mean values converge to the well known deterministic solution @xmath218 \end{aligned}\ ] ] with the common phase lag @xmath219 .
therefore , after deterministic transients have settled the _ cyclo - stationary _ probability density of the driven oscillator reads @xmath220 with the gaussian density @xmath221 } .\ ] ] using eq .
( [ eq : ricefreq ] ) the cyclo - stationary probability density ( [ e_pharm ] ) yields an oscillating expression for the rice frequency @xmath222 .
the time dependence of this stochastic average can be removed by an initial phase average , i.e. , a subsequent average over one external driving period @xmath223 , @xmath224 the resulting analytical and numerically achieved values of the rice frequency as a function of the noise intensity @xmath77 are shown in fig .
[ f_oscillatordriven ] for fixed @xmath225 and various values of @xmath114 . for small noise intensities @xmath77 the rice frequency @xmath176 is identical to the external driving frequency @xmath100 , whereas for large noise intensities the external drive becomes inessential and the rice frequency approaches @xmath211 .
further insight into the analytic expression ( [ e_drivenosc ] ) is gained from performing the following scale transformations @xmath226 from which we immediately find the rescaled velocity @xmath227 inserting these dimensionless quantities into eq .
( [ e_drivenosc ] ) yields @xmath228 i(\tilde a,\tilde{\omega}_0)&=&{1\over\pi } \int\limits_{-\delta}^{2\pi-\delta}\;\int\limits_0^{\infty } \tilde v\,\exp\big[\!\ ! -(\tilde v + \tilde a \sin\tilde t)^2 \nonumber\\[-.3 cm ] & & \qquad\qquad\qquad\qquad -(\tilde{\omega}_0\,\tilde a \cos\tilde t)^2 \big]\,d\tilde v\;d\tilde t \label{eq : i}\end{aligned}\ ] ] where we have defined further dimensionless quantities @xmath229 \tilde{\omega}_0 & = & \frac{{\omega}_0}{\omega } .\end{aligned}\ ] ] due to the @xmath24 periodicity of the trigonometric functions , the integral ( [ eq : i ] ) does not change when shifting the interval for the integration with respect to @xmath230 back to @xmath231 $ ] .
hence , @xmath232 is only a function of @xmath233 and @xmath234 .
an expansion for small @xmath233 yields @xmath235\ ] ] which implies for large @xmath236 @xmath237 the opposite extreme , @xmath238 or @xmath239 , can be extracted from a saddlepoint approximation around @xmath240 and @xmath241 . following this procedure ,
the integral ( [ eq : i ] ) gives the constant @xmath242 .
this directly implies @xmath243 .
the crossover between these two extremes occurs when the first correction term in ( [ eq : a20 ] ) is no longer negligible , i.e. , for @xmath244 when solved for the crossover noise intensity @xmath245 , this yields @xmath246}\,,\ ] ] which , for the parameters used in fig .
[ f_oscillatordriven ] , correctly gives values between @xmath247 and @xmath248 .
in fig . [ f_oscillatordriven ] the parameters @xmath249 , and @xmath212 and , hence , @xmath234 are identical for all curves . solving @xmath250 with respect
to @xmath251 shows that the curves become shifted horizontally as in the log - linear plot in fig .
[ f_oscillatordriven ] . another way to explain
this shift is by noting that @xmath252 .
the bistable kramers oscillator , i.e. , eq .
( [ e_osc ] ) with the double well potential @xmath253 is often used as a paradigm for nonlinear systems . with reference to eq .
( [ e_osc ] ) the corresponding langevin equation is given by @xmath254 which , in the absence of the external signal , @xmath147 , generates the stationary probability distribution @xmath255 with the normalization constant @xmath160 .
using this stationary probability density and eq .
( [ eq : ricefreq ] ) we can determine the rice frequency analytically . in fig .
[ f_bioscstatic ] we depict this analytic result together with numerical simulation data including error bars .
the simulation points perfectly match the analytically determined curve . as expected for the asymptotically dominant quartic term , i.e. , @xmath256 ( cf .
sec.[sec:3:general ] , especially eq .
( [ eq : largenoise ] ) ) , the rice frequency scales as @xmath257 for large values of @xmath77 . comparing the rice frequency formula , eq .
( [ eq : ricefreq ] ) , with the forward jumping rate @xmath258 from the transition state theory @xcite ,
@xmath259 \label{eq : tstint}\ ] ] where @xmath260\ , , \label{eq : semipf}\ ] ] and @xmath261 represents the corresponding hamiltonian , one can see that the difference between both solely rests upon normalizing prefactors .
whereas the rate @xmath258 is determined by the division of the integral eq .
( [ eq : tstint ] ) by the `` semipartition '' function @xmath262 , the rate @xmath263 is established by dividing the same integral eq .
( [ eq : tstint ] ) by the complete partition function @xmath264 @xmath265\ , .
\label{eq : fullpf}\ ] ] particularly for symmetric ( unbiased ) potentials , i.e. , @xmath266 , this amounts to the relation @xmath267 , hence , @xmath268 at weak noise , @xmath269 , this relation simplifies to @xmath270 \ ; , \label{eq : tstraprox}\ ] ] wherein @xmath271 denotes the barrier height and @xmath212 the angular frequency inside the well ( @xmath272 ) .
indeed , in the small - to - moderate regime of weak noise this estimate nicely predicts the exact rice frequency ( cf .
[ f_bioscstatic ] ) .
the periodically driven bistable kramers oscillator was the first model considered to explain the phenomenon of sr @xcite and it still serves as one of the major paradigms of sr @xcite . in its overdamped form
it was used to support experimental data ( from the schmitt trigger ) displaying the effect of stochastic frequency locking @xcite observed for sufficiently large , albeit subthreshold signal amplitudes , i.e. , for @xmath273 . from a numerical simulation of the overdamped kramers oscillator and computing the hilbert phase it was also found that noise - induced frequency locking for large signal amplitudes was accompanied by noise - induced phase coherence
, the latter implies a pronounced minimum of the effective phase diffusion coefficient @xmath274 \label{eq : deff1}\ ] ] occurring for optimal noise intensity .
based on a discrete model @xcite , analytic expressions for the frequency and phase diffusion coefficient were derived that correctly reflect the conditions for noise - induced phase synchronization @xcite for both periodic and aperiodic input signals . to link the mentioned results to the rice frequency introduced above we next investigate the behaviour of the kramers oscillator with non - vanishing inertia .
we show numerical simulations for eq .
( [ e_biosc ] ) with the parameters @xmath275 and diverse values of @xmath276 in fig .
[ f_nonlinearf ]
. for larger values of @xmath276 , a region around @xmath277 appears where the rice frequency is locked to the external driving frequency @xmath100 . since for larger values of the external driving @xmath276
smaller values of the noise parameter @xmath77 are needed to obtain the same rate for switching events , the entry into the locking region shifts to smaller values of @xmath77 for increasing @xmath276 . in fig .
[ f_nonlinearrice1 ] we present numerical simulations for fixed @xmath278 and different values of the damping coefficient @xmath114 .
note that the value of @xmath276 is slightly smaller than the critical value @xmath279 . for smaller values of @xmath114
wider coupling regions appear since it is easier for the particle to follow the external driving for smaller damping . to check whether frequency synchronization is accompanied by effective phase synchronization we have also computed the averaged effective phase diffusion coefficient , this time defined by the following asymptotic expression @xmath280 ^ 2 \right\rangle\,.\ ] ] it should be clear that the instantaneous `` rice '' phase @xmath3 was determined via zero crossings . the connection with the instantaneous diffusion coefficient defined in ( [ eq : deff1 ] ) is established by applying the limit @xmath281 @xmath282 in fig .
[ f_nonlineardiff1 ] we show numerical simulations of the effective phase diffusion coefficient @xmath283 as function of noise intensity @xmath77 .
the phase diffusion coefficient displays a local minimum that gets more pronounced if the damping coefficient @xmath114 is decreased .
indeed , phase synchronization reveals itself through this local minimum of the average phase diffusion coefficient @xmath283 in the very region of the noise intensity @xmath77 where we also observe frequency synchronization , cf .
[ f_nonlinearf ] .
the qualitative behaviour of the diffusion coefficient agrees also with a recently found result related to diffusion of brownian particles in biased periodic potentials @xcite .
a necessary condition for the occurrence of a minimum was an anharmonic potential in which the motion takes place . in this biased anharmonic potential
the motion over one period consists of a sequence of two events .
every escape over a barrier ( arrhenius - like activation ) is followed by a time scale induced by the bias and describing the relaxation to the next minimum .
the second step is weakly dependent on the noise intensity and the relaxation time may be even larger then the escape time as a result of the anharmonicity . for such potentials
the diffusion coefficient exhibits a minimum for optimal noise , similar to the one presented in figs .
[ f_nonlineardiff1 ] and [ f_nonlineardiff1a ] .
the average duration of locking episodes @xmath284 can be computed by equating the second moment of the phase difference ( between the driving signal and the oscillator ) to @xmath285 @xcite . a rough estimate , valid for the regions where frequency synchronization occurs
, i.e. , where the dynamics of the phase difference is dominated by diffusion , thus reads @xmath286 or , when expressed by the number of driving periods @xcite @xmath287 in this way we estimate from figs .
[ f_nonlineardiff1 ] and [ f_nonlineardiff1a ] @xmath288 for @xmath289 and relevant @xmath283 varying between @xmath290 . in the previous examples
we have shown how frequency synchronization , revealing itself through a plateau of the output frequency matching the harmonic input frequency , and reduced phase diffusivity together mark the occurrence of noise - enhanced phase coherence .
optimal noise intensities were found in the range where one also observes sr ( in the overdamped system ) . in order to underline that under certain conditions sr exists but may not be accompanied by effective phase synchronization we present simulation results for the rice frequency and the diffusion coefficient in fig .
[ fig : srnosync ] obtained for the bistable kramers oscillator with a friction coefficient @xmath145 and external frequency @xmath291 . for noise intensities
@xmath292 the output frequency matches @xmath100 and nearby the overdamped kramers oscillator exhibits the phenomenon of sr , i.e. , one finds a maximum of the spectral power amplification @xcite .
in contrast , we neither can find a minimum in the diffusion coefficient nor a plateau around @xmath100 meaning that no phase coherence and not even frequency synchronization can be observed .
the reason is that the external signal switches much too fast for the bistable system to follow ; note that in the two - state description with arrhenius rates the prefactor @xmath75 ( cf .
eq . [ a1a2 ] ) restricts the switching frequency from above .
noise - induced phase coherence requires a device with a faster internal dynamics , i.e. , @xmath293 .
we underline that the noise - induced phase synchronization is a much more stringent effect than stochastic resonance .
this statement becomes most obvious when recalling that the spectral power amplification attains a maximum at an optimal noise intensity for arbitrarily small signal amplitudes and any frequency of the external signal .
in contrast , noise - induced phase synchronization and even frequency locking are nonlinear effects and as such require amplitude and frequency to obey certain bounds ( see the `` arnold tongues '' in sec .
[ sec:2 ] ) .
we expect that the functioning of important natural devices , e.g. communication and information processing in neural systems or sub - threshold signal detection in biological receptors , rely on phase synchronization rather than stochastic resonance .
the authors acknowledge the support of this work by the deutsche forschungsgemeinschaft , sfb 555 `` komplexe nichtlineare prozesse '' , project a 4 ( l.s .- g . and j.f .
) , sfb 486 `` manipulation of matter on the nanoscale '' , project a 10 ( p.h . ) and deutsche forschungsgemeinschaft project ha 1517/13 - 4 ( p.h . ) .
10 v. i. arnold , trans . of the am .
soc . * 42 * , 213 ( 1965 ) ; e. ott , _ chaos in dynamical systems _ , ( cambridge university press , cambridge , 1993 ) .
stratonovich , _ topics in the theory of random noise _ , ( gordon and breach , new york , 1967 )
. h. fujisaka and t. yamada , prog .
phys.*69 * , 32 ( 1983 ) ; a.s .
pikovsky , z. phys .
b * 55 * , 149 ( 1984 ) ; l.m .
pecora and t.l .
carroll , phys .
rev . lett . * 64 * , 821 ( 1990 ) .
rulkov , m.m .
sushchik , l.s .
tsimring , and h.d.i .
abarbanel , phys .
e * 51 * , 980 ( 1995 ) ; l. kocarev and u. parlitz , phys .
lett * 76 * , 1816 ( 1996 ) .
rosenblum , a.s .
pikovsky , and j. kurths , phys .
lett . * 78 * , 4193 ( 1997 ) ; s. taherion and y.c .
lai , phys .
e * 59 * , r6247 ( 1999 ) ; m.g .
rosenblum , a.s .
pikovsky , and j. kurths , phys .
lett . * 76 * , 1804 ( 1996 ) .
han , t.g .
yim , d.e .
postnov , and o.v .
sosnovtseva , phys .
83 * , 1771 ( 1999 ) .
izhikevich , int .
chaos * 10 * , 1171 ( 2000 ) ; bambi hu and changsong zhou , phys .
e * 63 * , 026201 ( 2001 ) .
v. anishchenko , a. neiman , a. astakhov , t. vadiavasova , and l. schimansky - geier , _ chaotic and stochastic processes in dynamic systems _ , ( springer , berlin , 2002 ) .
l. schimansky - geier , v. anishchenko , and a. neiman , in _ neuro - informatics _ , s. gielen and f. moss ( eds . ) , _ handbook of biological physics _ , vol .
4 , series editor a.j . hoff , ( elsevier science , 2001 ) .
a. neiman , x. pei , d. russell , w. wojtenek , l. wilkens , f. moss , h.a .
braun , m.t .
huber , and k. voigt , phys .
* 82 * , 660 ( 1999 ) , s. coombes and p.c .
bressloff , phys .
e * 60 * , 2086 ( 1999 ) and phys .
e * 63 * , 059901 ( 2001 ) ; w. singer , neuron * 24 * , 49 ( 1999 ) ; r.c .
elson , a.i .
selverston , r. huerta , n.f .
rulkov , m.i .
rabinovich , and h.d.i .
abarbanel , phys .
lett . * 81 * , 5692 ( 1998 ) ; p. tass , m.g .
rosenblum , j. weule , j. kurths , a. pikovsky , j. volkmann , a. schnitzler , and h .- j .
freund , phys .
* 81 * , 3291 ( 1998 ) ; r. ritz and t.j .
sejnowski , current opinion in neurobiology * 7 * , 536 ( 1997 ) .
b. schfer , m.g .
rosenblum , and j. kurths , nature ( london ) * 392 * , 239 ( 1998 ) .
b. blasius , a. huppert , and l. stone , nature ( london ) * 399 * , 354 ( 1999 ) .
j.p.m . heald and j. stark , phys .
lett . * 84 * , 2366 ( 2000 ) .
freund , a.b .
neiman , and l. schimansky - geier , europhys . lett . *
50 * , 8 ( 2000 ) .
l. callenbach , p. hnggi , s.j .
linz , j.a .
freund , l. schimansky - geier , phys .
e ( in press ) .
s.o . rice , bell system technical j. , * 23 /24 * , 1 - 162 ( 1944/1945 ) ; note sec .
3.3 on p. 57 - 63 in this double volume therein .
s.o . rice , in _ selected papers on noise and stochastic processes _ , n. wax , ed .
189 - 195 therein , ( dover , new york , 1954 ) l. gammaitoni , p. hnggi , p. jung , and f. marchesoni , rev .
mod . phys . * 70 * , 223 ( 1998 ) .
anishchenko , a.b .
neiman , f. moss , and l. schimansky - geier , phys .
usp . * 42 * , 7 ( 1999 ) .
a. neiman , a. silchenko , v. anishchenko , and l. schimansky - geier , phys .
e * 58 * , 7118 ( 1998 ) .
p. jung , phys .
rep . * 234 * , 175 ( 1995 ) .
n.n . bogoliubov and y.a .
mitropolski , _ asymptotic methods in the theory of non - linear oscillations _ , ( gordon and breach sciences publishers , new york , 1961 ) .
the explicit expression for the phase involving the @xmath294 has to be understood in the sense of adding multiples of @xmath59 to make it a continuously growing function of time .
p. hnggi and p. riseborough , am .
* 51 * , 347 ( 1983 ) .
the statement requires the function @xmath8 defined in eq .
( [ eq : bogmist1b ] ) to remain finite for @xmath295 ; this , however , is no severe restriction .
winfree , j. theor .
biol . , * 16 * 15 ( 1967 ) ; j. guckenheimer , j. math . biol . * 1 * , 259 ( 1975 ) .
d. gabor , j. iee ( london ) * 93 * ( iii ) , 429 ( 1946 ) .
a. pikovsky , m. rosenblum , and j. kurths , _ synchronization : a universal concept in nonlinear sciences _ , ( cambridge university press , cambridge , 2001 ) .
vainstein and d.e .
vakman , _ frequency analysis in the theory of oscillations and waves _ ( in russian ) , ( nauka , moscow , 1983 ) .
r. carmona , w .- l .
hwang , b. torresani , _ practical time - frequency analysis _ , ( academic press , san diego , 1998 ) .
lachaux , e. rodriguez , m. le van quen , a. lutz , j. martinerie , and f. varela , int .
chaos * 10 * , 2429 ( 2000 ) ; m. le van quen , j. foucher , j .-
lachaux , e. rodriguez , a. lutz , j. martinerie , and f. varela , j. neurosci . meth . *
111 * , 83 ( 2001 ) .
deshazer , r. breban , e. ott , and r. roy , phys .
* 87 * , 044101 ( 2001 ) .
b. mcnamara and k. wiesenfeld , phys .
a * 39 * , 4854 ( 1989 ) .
cox , _ renewal theory _ ,
( chapman & hall , london , 1967 ) . j.a .
freund , a.b .
neiman , and l. schimansky - geier , in p. imkeller and j. von storch ( eds . ) , stochastic climate models .
progress in probability , ( birkhuser , boston , basel , 2001 ) .
n. g. van kampen , _
stochastic processes in physics and chemistry _ rev . and
( north - holland , amsterdam , 1992 ) .
r. adler , proc .
ire * 34 * , 351 ( 1946 ) ; p. hnggi and p. riseborough , am . j. phys .
* 51 * , 347 ( 1983 ) . c. van den broeck , phys
e * 47 * , 4579 ( 1993 ) ; u. zrcher and c. r. doering , phys .
e * 47 * , 3862 ( 1993 ) .
f. marchesoni , f. apostolico , and s. santucci , phys .
e * 59 * , 3958 ( 1999 ) .
a. neiman , l. schimansky - geier , f. moss , b. shulgin , and j. j. collins , phys .
e * 60 * , 284 ( 1999 ) .
b. shulgin , a. neiman , and v. anishchenko , phys .
* 75 * , 4157 ( 1995 ) .
r. rozenfeld , j.a .
freund , a. neiman , l. schimansky - geier , phys .
e 64 , 051107 ( 2001 ) .
p. hnggi , p. talkner , and m. borkovec , rev .
* 62 * , 251 ( 1990 ) .
p. hnggi and h. thomas , phys
. rep . * 88 * , 207 ( 1982 ) ; see sec . 2.4 therein .
p. hnggi and p. jung , adv .
* 89 * , 239 ( 1995 ) ; p. hnggi , f. marchesoni , and p. grigolini , z. physik b*56 * , 333 ( 1984 ) .
r. benzi , g. parisi , and a. vulpiani , j. phys .
a : math gen . * 14 * , l453 ( 1981 ) ; c. nicolis , solar physics * 74 * , 473 ( 1981 ) .
a. neiman , l. schimansky - geier , f. moss , b. shulgin , and j. j. collins , phys .
e * 60 * , 284 ( 1999 ) .
b. lindner , m. kostur and l. schimansky - geier , fluct . and
noise letters * 1 * , r25 ( 2001 ) .
freund , j. kienert , l. schimansky - geier , b. beisner , a. neiman , d.f .
russell , t. yakusheva , and f. moss , phys .
e * 63 * , 031910 ( 2001 ) . | the phenomenon of frequency and phase synchronization in stochastic systems requires a revision of concepts originally phrased in the context of purely deterministic systems .
various definitions of an instantaneous phase are presented and compared with each other with special attention payed to their robustness with respect to noise .
we review the results of an analytic approach describing noise - induced phase synchronization in a thermal two - state system . in this context
exact expressions for the mean frequency and the phase diffusivity are obtained that together determine the average length of locking episodes . a recently proposed method to quantify frequency synchronization in noisy potential systems
is presented and exemplified by applying it to the periodically driven noisy harmonic oscillator . since this method is based on a threshold crossing rate pioneered by s.o .
rice the related phase velocity is termed rice frequency .
finally , we discuss the relation between the phenomenon of stochastic resonance and noise - enhanced phase coherence by applying the developed concepts to the periodically driven bistable kramers oscillator . *
studying synchronization phenomena in stochastic systems necessitates a revision of concepts originally developed for deterministic dynamics .
this statement becomes obvious when considering the famous phase - locking effect @xcite : unbounded fluctuations that occur , for instance , in gaussian noise will always prevent the existence of a strict bound for the asymptotic phase difference of two systems .
nevertheless , a reformulation of the synchronization phenomenon in the presence of noise is possible by quantifying the average duration @xmath0 of locking epochs that are disrupted by phase slips . in case that @xmath1 , where @xmath2 is some characteristic time of the dynamics , e.g. the period of an external drive or the inverse of some intrinsic natural frequency , it is justified to speak about effective phase synchronization . * |
classic optimization settings assume that the problem data are known exactly .
robust optimization , like stochastic optimization , instead assumes some degree of uncertainty in the problem formulation .
most approaches in robust optimization formalize this uncertainty by assuming that all uncertain parameters @xmath0 are described by a set of possible outcomes @xmath1 , the uncertainty set . for general overviews on robust optimization , we refer to @xcite .
while the discussion of properties of the robust problem for different types of uncertainty sets @xmath1 has always played a major role in the research community , only recently the data - driven design of useful sets @xmath1 has become a focus of research . in @xcite ,
the authors discuss the design of @xmath1 taking problem tractability and probabilistic guarantees of feasibility into account .
the paper @xcite discusses the relationship between risk measures and uncertainty sets . in distributionally robust optimization
, one assumes that a probability distribution on the data is roughly known ; however , this distribution itself is subject to an uncertainty set @xmath1 of possible outcomes ( see @xcite ) .
another related approach is the globalized robust counterpart , see @xcite .
the idea of this approach is that a relaxed feasibility should be maintained , even if a scenario occurs that is not specified in the uncertainty set .
the larger the distance of @xmath0 to @xmath1 , the further relaxed becomes the feasibility requirement of the robust solution . in this work we present an alternative to constructing a specific uncertainty set @xmath1 .
instead , we only assume knowledge of a nominal ( undisturbed ) scenario , and consider a set of possible uncertainty sets of varying size based on this scenario
. that is , a decision maker does not need to determine the size of uncertainty ( a task that is usually outside his expertise ) .
our approach constructs a solution for which the worst - case objective with respect to any possible uncertainty set performs well on average over all uncertainty sizes . the basic idea of variable - sized uncertainty
was recently introduced in @xcite .
there , the aim is to construct a set of robust candidate solutions that requires the decision maker to chose one that suits him best . in our setting
, we consider all uncertainty sizes simultaneously , and generate a single solution as a compromise approach to the unknown uncertainty .
we call this setting the _ compromise approach to variable - sized uncertainty_. we focus on combinatorial optimization problems with uncertainty in the objective function , and consider both min - max and min - max regret robustness ( see @xcite ) .
this work is structured as follows . in section [ sec : var ] , we briefly formalize the setting of variable - sized uncertainty .
we then introduce our new compromise approach for min - max robustness in section [ sec : comp ] , and for the more involved case of min - max regret robustness in section [ sec : comp2 ] .
we present complexity results for the selection problem , the minimum spanning tree problem , and the shortest path problem in section [ sec : probs ] . in section [ sec : exp ] , we evaluate our approach in a computation experiment , before concluding this paper in section [ sec : conc ] .
we briefly summarize the setting of @xcite .
consider an uncertain combinatorial problem of the form @xmath2 with @xmath3 , and an uncertainty set @xmath4 that is parameterized by some size @xmath5 . for example , * interval - based uncertainty @xmath6 } [ ( 1-{\lambda})\hat{c}_i,(1+{\lambda})\hat{c}_i]$ ] with @xmath7 $ ] , or * ellipsoidal uncertainty @xmath8 with @xmath9 .
we call @xmath10 the _ nominal scenario _ , and any @xmath11 that is a minimizer of p(@xmath10 ) a _ nominal solution_. in variable - sized uncertainty , we want to find a set of solutions @xmath12 that contains an optimal solution to each robust problems over all @xmath13 . here ,
the robust problem is either given by the min - max counterpart @xmath14 or the min - max regret counterpart @xmath15 in the case of min - max robustness , such a set can be found through methods from multi - objective optimization in @xmath16 , where @xmath17 denotes the complexity of the nominal problem , for many reasonable uncertainty sets
. however , @xmath18 may be exponentially large . furthermore , in some settings , a set of solutions that would require a decision maker to make the final choice may not be desirable , but instead a single solution that represents a good trade - off over all values for @xmath13 is sought .
in this paper we are interested in finding one single solution that performs well over all possible uncertainty sizes @xmath5 . to this end , we consider the problem @xmath19 for some weight function @xmath20 , such that @xmath21 is well - defined .
we call this problem the compromise approach to variable - sized uncertainty .
the weight function @xmath22 can be used to include decision maker preferences ; e.g. , it is possible to give more weight to smaller disturbances and less to large disturbances . if a probability distribution over the uncertainty size were known , it could be used to determine @xmath22 . in the following , we consider ( c ) for different shapes of @xmath23 .
[ th1 ] let @xmath6 } [ ( 1-{\lambda})\hat{c}_i,(1+{\lambda})\hat{c}_i]$ ] be an interval - based uncertainty set with @xmath24 $ ] . then , a nominal solution @xmath25 is an optimal solution of ( c ) . as @xmath26
, we get @xmath27 therefore , a minimizer of the nominal problem with costs @xmath10 is also a minimizer of ( c ) .
[ lem1 ] for an ellipsoidal uncertainty set @xmath28 with @xmath29 , it holds that @xmath30 this result has been shown in @xcite for @xmath31 .
the proof holds analogously .
[ th2 ] let @xmath32 be an ellipsoidal uncertainty set with @xmath33 .
then , an optimal solution to ( c ) can be found by solving a single robust problem with ellipsoidal uncertainty .
using lemma [ lem1 ] , we find @xmath34 to find a minimizer of @xmath21 , we can therefore solve the robust counterpart of ( p ) using an uncertainty set @xmath35 with @xmath36 .
note that @xmath37 , if @xmath38 , i.e. , the compromise solution simply hedges against the average size of the uncertainty .
in general , recall that this formula gives the centroid of the curve defined by @xmath22 . the results of theorems [ th1 ] and [ th2 ] show that compromise solutions are easy to compute , as the resulting problems have a simple structure .
this is due to the linearity of the robust objective value in the uncertainty size @xmath13 .
such linearity does not exist for min - max regret , as is discussed in the following section .
we now consider the compromise approach in the min - max regret setting . in classic min - max regret ,
one considers the problem @xmath39 with @xmath40 . in the following ,
we restrict the analysis to the better - researched interval uncertainty sets @xmath6 } [ ( 1-{\lambda})\hat{c}_i,(1+{\lambda})\hat{c}_i]$ ] .
ellipsoidal uncertainty have been introduced in min - max regret only recently ( see @xcite ) .
the compromise approach to variable - sized uncertainty becomes @xmath41 to simplify the presentation , we assume @xmath42 $ ] and @xmath38 for all @xmath43 in the following .
all results can be directly extended to piecewise linear functions @xmath22 with polynomially many changepoints .
we first discuss the objective function @xmath21 for some fixed @xmath44 .
note that @xmath45 } \hat{c}_i ( 1-{\lambda}+ 2{\lambda}x_i ) y_i.\ ] ] hence , @xmath46 is a piecewise linear function in @xmath13 , where every possible regret solution @xmath47 defines an affine linear regret function @xmath48 , with @xmath49 .,scaledwidth=60.0% ] figure [ fig1 ] illustrates the objective function . in red is the maximum over all regret functions , which defines @xmath21 . on the interval
@xmath50 $ ] , the regret of some solution @xmath51 is defined through @xmath52 , while solution @xmath53 defines the regret on @xmath54 $ ] , and @xmath55 defines the regret on @xmath56 $ ] . in this case , we can hence compute @xmath57 in general , to compute @xmath21 , we need to determine all relevant regret solutions @xmath47 , and the intersections of the resulting regret functions .
let @xmath58 be the set of changepoints of the piecewise linear function @xmath59 . to formulate problem ( c ) as a linear integer program
, we use a set @xmath60 . as
@xmath61 is a finite set , there always exists a set @xmath62 that is finite .
in general , it may contain exponentially many elements .
for the ease of notation , we assume @xmath63 with @xmath64 and @xmath65 . as a first approach
, we model @xmath21 by using all possible regret solutions @xmath66
. @xmath67 }
( 1+{\overline{\lambda}}_j)\hat{c}_i x_i - \sum_{i\in[n ] } ( 1-{\overline{\lambda}}_j + 2{\overline{\lambda}}_jx_i)\hat{c}_i y^\ell_i & \forall \lambda_j \in{\overline{\lambda } } , y^\ell\in{\mathcal{x}}\nonumber\\ & x\in{\mathcal{x}}\nonumber\end{aligned}\ ] ] where @xmath68 .
if ( p ) has a duality gap of zero , i.e. , if solving the dual of the linear relaxation also gives an optimal solution to ( p ) , this formulation can be simplified .
examples where this is the case include the shortest path problem , or the minimum spanning tree problem .
let us assume that the linear relaxation of @xmath61 is given by @xmath69 for the regret problem @xmath70 we may then write the following equivalent problem ( see @xcite ) : @xmath71 using this reformulation , we find the following program for ( c ) @xmath72 for binary variables @xmath51 , the product @xmath73 can then be linearized .
if a set @xmath62 can be found that is of polynomial size , this is a compact formulation .
in general , constraints and variables can be added in an iterative algorithm that generates new candidate values for @xmath74 in problem .
if the zero duality gap assumption does not hold , we can use formulation , where both values for @xmath13 and regret solutions @xmath47 need to be generated .
this approach is explained in the following section . in the following
we describe how to compute the set @xmath75 of changepoints of @xmath59 .
this is then used to solve formulation in the case of a zero duality gap for ( p ) as described in algorithm [ alg : c ] .
@xmath76 , @xmath77 [ a1s1 ] solve formulation using @xmath62 .
let the solution be @xmath78 , and the objective value @xmath79.[algret ] compute @xmath80 .
let the resulting changepoints be @xmath81 , and the objective value @xmath82 .
[ a1s3 ] * end * : @xmath78 is an optimal solution . [ a1s5 ] @xmath83 [ a1s7 ] @xmath84 go to [ algret ] the algorithm begins with a starting set @xmath85 as a guess for relevant changepoints ( any other set could be used here ) . using the current set @xmath62 , it then solves formulation .
as not all constraints of the problem are present , this is a problem relaxation .
accordingly , the objective value that is found is only a lower bound @xmath79 on the true optimal objective value of the problem . to evaluate the resulting candidate solution
@xmath78 , we compute @xmath80 in step [ a1s3 ] . the sub - algorithm for this purpose
is explained below .
as @xmath78 is a feasible solution to ( c ) , @xmath80 gives an upper bound @xmath82 on the optimal objective value .
hence , if lower and upper bound coincide , an optimal solution has been found . otherwise , we extend the set @xmath62 in step [ a1s7 ] and repeat the procedure . if ( p ) has a duality gap larger than zero , the same algorithm can be used with the slight adjustment that problem is solved in step [ algret ] . to this end , also regret solutions @xmath86 generated in the computation of @xmath80 need to be collected .
we describe the procedure to compute @xmath21 in algorithm [ alg : val ] .
@xmath87 , @xmath88 @xmath89 [ j1 ] solve ( p ) with costs @xmath90 .
let @xmath47 be the resulting solution .
@xmath91 @xmath92 change @xmath93 false calculate @xmath74 as the point where the affine linear regret functions defined by @xmath94 and @xmath95 intersect .
@xmath97 change @xmath93 true go to [ j1 ] reduce @xmath75 and @xmath98 such that only changepoints and regret functions remain that define the maximum over all affine linear functions .
@xmath75 , @xmath98 we begin with only two regret functions , for the extreme points @xmath99 .
the resulting two regret functions will intersect at one new candidate changepoint @xmath74 .
we find the regret solution @xmath47 maximizing the regret at this point by solving a problem of type ( p ) .
we then repeat this process by iteratively calculating the regret at all current intersection points .
note that if there exists any regret function that is larger than all current regret functions at some point @xmath74 , then it is also larger than all current functions at the intersection point between two of them .
hence , algorithm [ alg : val ] finds all relevant changepoints @xmath13 .
as it may also produce unnecessary candidates @xmath13 , we reduce the solution sets at the end to contain only those changepoints and regret functions that define the maximum .
the minimum selection problem is given by @xmath100 } x_i = p,\ x\in\{0,1\}^n\right\}\ ] ] and has been frequently studied in the literature on min - max regret optimization ( see @xcite ) . the min - max regret problem can be solved in @xmath101 time , see @xcite .
we show that also problem ( c ) can be solved in polynomial time .
let @xmath102 } [ ( 1-{\lambda})\hat{c}_i,(1+{\lambda})\hat{c}_i]$ ] for a fixed @xmath13 .
then @xmath25 is an optimal solution to the min - max regret selection problem .
we assume that items are sorted with respect to @xmath10 .
let @xmath103 be an optimal solution with @xmath104 for an item @xmath105 .
we assume @xmath106 is the smallest such item .
then there exists some @xmath107 with @xmath108 .
consider the solution @xmath109 with @xmath110 for @xmath111 and @xmath112 , @xmath113 .
let @xmath114 be a regret solution for @xmath103 .
we can assume that @xmath115 , as @xmath116 must be one of the @xmath117 cheapest items .
we can also assume @xmath118 , as @xmath119 is not among the @xmath117 cheapest items .
let @xmath120 be the regret solution for @xmath109 .
the solutions @xmath103 and @xmath109 differ only on the two items @xmath106 and @xmath121 .
hence , the following cases are possible : * case @xmath122 and @xmath123 , i.e. , @xmath124
. we have @xmath125 * case @xmath126 and @xmath127 , @xmath128 for some @xmath129 with @xmath130 .
note that this means @xmath131 , as otherwise , the regret solution of @xmath114 could be improved .
hence , @xmath132 * case @xmath133 and @xmath134 for some @xmath135 with @xmath136 . as the costs of item @xmath106 have increased by using solution @xmath109 instead of @xmath103 ,
the resulting two cases are analogue to the two cases above .
overall , solution @xmath109 has regret less or equal to the regret of @xmath103 . note that this result does not hold for general interval uncertainty sets , where the problem is np - hard .
it also does not necessarily hold for other combinatorial optimization problems ; e.g. , a counter - example for the assignment problem can be found in @xcite . finally , it remains to show that @xmath21 can also be computed in polynomial time .
for the compromise min - max regret problem of minimum selection it holds that @xmath137 for any fixed @xmath44 , and there is a set @xmath62 with @xmath138 . if @xmath51 is fixed , then there are @xmath117 items @xmath106 with costs @xmath139 , and @xmath140 items @xmath106 with costs @xmath141
. the regret solution is determined by the @xmath117 smallest items .
accordingly , when @xmath13 increases , the regret solution only changes if an item @xmath106 with @xmath142 , that used to be among the @xmath117 smallest items , moves to the @xmath140 largest items , and another item @xmath121 with @xmath143 becomes part of the @xmath117 smallest items .
there are at most @xmath144 values for @xmath13 where this is the case .
we define @xmath62 to consist of all @xmath145 $ ] such that @xmath146 for some @xmath147 $ ] , as only for such values of @xmath13 an optimal regret solution may change . hence , @xmath138 . as the size of @xmath75
is polynomially bounded , @xmath21 can be computed in polynomial time , and we get the following conclusion .
the compromise min - max regret problem of minimum selection can be solved in polynomial time .
the min - max regret spanning tree problem in a graph @xmath148 has previously been considered , e.g. , in @xcite .
the regret of a fixed solution can be computed in polynomial time , but it is np - hard to find an optimal solution .
we now consider the compromise min - max regret counterpart ( c ) .
let any spanning tree @xmath51 be fixed . to compute @xmath21 , we begin with @xmath149 and calculate a regret spanning tree by solving a nominal problem with costs @xmath10 . recall that this can be done using kruskal s algorithm that considers edges successively according to an increasing sorting @xmath150 with respect to costs .
if @xmath74 increases , edges that are included in @xmath51 have costs @xmath151 ( i.e. , their costs increase ) and edges not in @xmath51 have costs @xmath152 ( i.e. , their costs decrease ) .
kruskal s algorithm will only find a different solution if the sorting of edges change .
as there are @xmath153 edges with increasing costs , and @xmath154 edges with increasing costs , the sorting can change at most @xmath155 times ( note that two edges with increasing costs or two edges with decreasing costs can not change relative positions ) .
we have therefore shown : a solution to the compromise min - max regret problem of minimum spanning tree can be evaluated in polynomial time .
if the solution @xmath51 is not known , we can still construct a set @xmath62 with size @xmath156 that contains all possible changepoints along the same principle .
we can conclude : there exists a compact mixed - integer programming formulation for the compromise min - max regret problem of minimum spanning tree .
however , we show in the following that solving the compromise problem is np - hard . to this end
, we use the following result : @xcite the min - max regret spanning tree problem is np - hard , even if all intervals of uncertainty are equal to @xmath157 $ ] . note that if all intervals are of the form @xmath158 $ ] , then @xmath159 therefore , the min - max regret problem with costs @xmath157 $ ] is equivalent to the min - max regret problem with any other costs @xmath158 $ ] , in the sense that objective values only differ by a constant factor and both problems have the same set of optimal solutions . in particular , a solution @xmath47 that maximizes the regret of @xmath51 with respect to cost intervals @xmath158 $ ] is also a maximizer of the regret for any other cost intervals @xmath160 $ ] .
we can conclude : the compromise problem of min - max regret minimum spanning tree is np - hard , even if @xmath38 for all @xmath145 $ ] .
let an instance of the min - max regret spanning tree problem with cost intervals @xmath157 $ ] be given .
consider an instance of the compromise problem with @xmath161 for all @xmath162 , and @xmath38 .
then @xmath163 hence , any minimizer of @xmath21 is also an optimal solution to the min - max regret spanning tree problem .
for the shortest path problem , we consider @xmath61 as the set of all simple @xmath164 paths in a graph @xmath148 ( for the min - max regret problem , see , e.g. , @xcite ) . as for the minimum spanning tree problem ,
the regret of a fixed solution can be computed in polynomial time , but it is np - hard to find an optimal solution . for the compromise problem ( c ) , we have : @xmath165 we can interpret the minimization problem as a weighted sum of the bicriteria problem @xmath166 the number of solutions we need to generate to compute @xmath21 can therefore be bounded by the number of solutions we can find through such weighted sum computations ( the set of extreme efficient solutions @xmath167 ) . for the compromise shortest path problem , it holds that @xmath168 . depending on the graph @xmath169
, the following bounds on the number of extreme efficient solutions @xmath167 ( see , e.g. , @xcite ) can be taken from the literature @xcite : * for series - parallel graphs , @xmath170 * for layered graphs with width @xmath22 and length @xmath171 , @xmath172 * for acyclic graphs , @xmath173 * for general graphs @xmath174 we can conclude : a solution to the compromise min - max regret problem of shortest path can be evaluated in polynomial time on series - parallel graphs and layered graphs with fixed width or length . note that the number of extreme efficient solutions is only an upper bound on @xmath75 . unfortunately , we can not hope to find a better performance than this bound , as the following result demonstrates . for any bicriteria shortest path instance with costs @xmath175 , @xmath176 for all @xmath162
, there is an instance of ( c ) and a solution @xmath51 where @xmath177 .
let an instance of the bicriteria shortest path problem be given , i.e. , a directed graph @xmath148 with arc costs @xmath178 for all @xmath162 .
as @xmath179 for all @xmath162 , we can assume w.l.o.g . that @xmath180 for all @xmath162 .
we create the following instance of ( c ) .
every arc @xmath181 is substituted by three arcs @xmath182 , @xmath183 and @xmath184 .
we set @xmath185 , @xmath186 and @xmath187 ( see figure [ fignp ] for an example of such a transformation ) .
let @xmath188 , @xmath189 and @xmath190 contain all edges of the respective type .
additionally , we choose an arbitrary order of edges @xmath191 , and create arcs @xmath192 .
we set costs of these arcs to be a sufficiently large constant @xmath193 .
finally , let @xmath51 be the path that follows all edges in @xmath194 and @xmath189 .
note that edges in @xmath190 can be contracted , but are shown for better readability .
+ note that @xmath193 is sufficiently large so that no regret path @xmath47 will use any edge in @xmath194 .
hence , if @xmath47 uses an edge in @xmath188 , it will also have to use the following edges in @xmath189 and @xmath190 , i.e. , @xmath47 corresponds to a path in the original graph @xmath148 .
the regret of @xmath51 is @xmath195 therefore , if @xmath13 goes from @xmath196 to @xmath197 , all extreme efficient paths in the original graph @xmath169 are used to calculate @xmath46 .
we now consider the complexity of finding a solution @xmath44 that minimizes @xmath21 .
note that the reduction in @xcite uses interval costs of the form @xmath157 $ ] and @xmath198 $ ] , which does not fit into our cost framework @xmath199 $ ] .
instead , we make use of the following result : @xcite the min - max regret shortest path problem is np - hard for layered graphs with interval costs @xmath157 $ ] . note that for layered graphs , all paths have the same cardinality .
hence , @xmath200 ( see section [ sec : mst ] ) , and the problem with costs @xmath157 $ ] is equivalent to the problem with costs @xmath158 $ ] for any @xmath201 .
analog to the last section , we can therefore conclude : finding an optimal solution to the compromise shortest path problem is np - hard on layered graphs , even if @xmath202 for all @xmath145 $ ] .
in this section we present two experiments on compromise solutions to min - max regret problems with variable - sized uncertainty .
the first experiment is concerned with the computational effort to find such a solution using the iterative algorithms presented in section [ sec : algo ] . in the second experiment
we compare these solutions to the alternatives of using classic regret solutions for various uncertainty set sizes .
both experiments are conducted on shortest path instances of two types .
the first type consists of complete layered graphs .
we parameterize such instances by the number of layers , width and cost types .
each graph consists of a source node , a sink node , and node layers of equal width between . for @xmath203 layers of width @xmath204 ,
an instance has a total of @xmath205 nodes and @xmath206 edges .
we use @xmath207 to @xmath208 in steps of size 5 , and @xmath209 .
graph sizes thus vary from 32 nodes and 130 edges to 1,122 nodes and 22,040 edges .
edges connect all nodes of one layer to the nodes of the next layer . source and sink
are completely connected to the first and last layer , respectively .
we considered two types of cost structures to generate @xmath10 . for type
a , all costs are chosen uniformly from the interval @xmath210 $ ] . for type b costs ,
we generate nominal costs in @xmath211\cup[70,100]$ ] , i.e. , they are either low or high . in total , there are @xmath212 parameter combinations . for each combination ,
we generate 20 instances , i.e. , a total of 1,760 instances .
the second type consists of graphs with two paths , that are linked by diagonal edges .
for some length parameter @xmath213 , we generated two separate paths from a node @xmath214 to a node @xmath215 , each with @xmath213 nodes between .
we then generate diagonal edges in the following way . on one of the two paths , we choose the @xmath106th node uniformly randomly .
we then connect this node with the @xmath121th node on the other path , where @xmath216 .
the @xmath121th node is chosen with probability @xmath217 , i.e. , long diagonal edges are unlikely ( ensuring that @xmath121 is at most @xmath213 )
. edges along the two base paths have length chosen uniformly from the interval @xmath210 $ ] . for diagonal edges ,
we determine their length by sampling from the same interval @xmath218 times , and adding these values , i.e. , all paths have the same expected length .
we generate instances with length @xmath213 from 50 to 850 in steps of 100 , and set the number of diagonal edges to be @xmath219 for @xmath220 .
the smallest instances therefore contain 102 nodes and 105 edges ; the largest instances contain 1,702 nodes and 1,830 edges . for each parameter combination ,
we generate 20 instances , i.e. , a total of @xmath221 instances .
the classic min - max regret shortest path problem on instances of both types is known to be np - hard , see @xcite .
we investigate both types , as we expect the nominal solution to show a different performance : for layered graphs , the nominal solution is also optimal for @xmath222 , as for every path there also exists a disjoint path .
therefore , the regret of a path @xmath223 with respect to @xmath222 is @xmath224 . for the second type of instances , a good solution with respect to min - max regret
can be expected to intersect with as many other paths as possible .
we can therefore expect the nominal solution to be different to the optimal solution of @xmath222 .
all experiments were conducted using one core of a computer with an intel xeon e5 - 2670 processor , running at 2.60 ghz with 20 mb cache , with ubuntu 12.04 and cplex v.12.6 .
we solve the compromise approach to variable - sized uncertainty for each instance using the algorithms described in section [ sec : algo ] and record the computation times .
average computation times in seconds are presented in table [ tab1 ] . in each column
we average over all instances for which this parameter is fixed ; i.e. , in column width 5 we show the results over all 440 instances that have a width of 5 , broken down into classes of different length .
the results indicate that computation times are still reasonable given the complexity of the problem , and mostly depend on the size of the instance ( width parameter ) and the density of the graph , while the cost structure has no significant impact on computation times .
.average computation times to solve ( c ) in seconds . [ cols= " > ,
> , > , > , > , > , > , > " , ] in our second experiment , we compare the compromise solution to the nominal solution ( which is also the min - max regret solution with respect to the uncertainty sets @xmath225 and @xmath222 ) , and to the min - max regret solutions with respect to @xmath226 , @xmath227 and @xmath228 .
to compare solutions , we calculate the regret of the compromise solution for values of @xmath13 in @xmath157 $ ] .
we take this regret as the baseline . for all other solutions
, we also calculate the regret depending on @xmath13 , and compute the difference to the baseline .
we then compute the average differences for fixed @xmath13 over all instances of the same size .
the resulting average differences are shown in figure [ fig : exp ] for four instance sizes . to set the differences in perspective , the average regret ranges from @xmath225 to @xmath222 of the compromise solutions
are shown in the captions .
+ by construction , a min - max regret solution with respect to @xmath229 has the smallest regret for this @xmath230 .
generally , all presented solutions have higher regret than the nominal solution for small and for large values of @xmath13 , and perform better in between . by construction
, the compromise solution has the smallest integral under the shown curve .
it can be seen that it presents an interesting alternative to the other solutions by having a relatively small regret for small and large values of @xmath13 , but also a relatively good performance in between .
we generate the same plots as in section [ sec : plots ] using the two - path instances .
recall that in this case , the nominal solution is not necessarily an optimal solution with respect to @xmath222 .
we therefore include an additional line for @xmath222 in figure [ fig : exp2 ] .
+ it can be seen that the nominal solution performs different to the last experiment ; the regret increases with @xmath74 in a rate that part of the line needed to be cut off from the plot for better readability .
the solution to @xmath227 performs very close to the compromise solution overall .
additionally , the scale of the plots show that differences in regret are much larger than in the previous experiment .
overall , it can be seen that using a robust solution plays a more significant role than in the previous experiment , as the nominal solution shows poor performance .
the solutions that hedge against large uncertainty sets ( @xmath228 and @xmath231 ) are relatively expensive for small uncertainty sets and vice versa .
the compromise solution ( as @xmath227 , in this case ) presents a reasonable trade - off over all uncertainty sizes .
classic robust optimization approaches assume that the uncertainty set @xmath1 is part of the input , i.e. , it is produced using some expert knowledge in a previous step .
if the modeler has access to a large set of data , it is possible to follow recently developed data - driven approaches to design a suitable set @xmath1 . in our approach , we remove the necessity of defining @xmath1 by using a single nominal scenario , and considering all uncertainty sets generated by deviating coefficients of different size simultaneously .
the aim of the compromise approach is to find a single solution that performs well on average in the robust sense over all possible uncertainty set sizes . for min - max combinatorial problems
, we showed that our approach can be reduced to solving a classic robust problem of particular size .
the setting is more involved for min - max regret problems , where the regret objective is a piecewise linear function in the uncertainty size .
we presented a general solution algorithm for this problem , which is based on a reduced master problem , and the iterative solution of subproblems of nominal structure . for specific problems ,
positive and negative complexity results were demonstrated .
the compromise selection problem can be solved in polynomial time .
solutions to the compromise minimum spanning tree problem can be evaluated in polynomial time , but it is np - hard to find an optimal solution .
for compromise shortest path problems , the same results hold in case of layered graphs ; however , for general graphs , it is still an open problem if there exist instances where exponentially many regret solutions are involved in the evaluation problem . in computational experiments we highlighted the value of our approach in comparison with different min - max regret solutions , and showed that computation times can be within few minutes for instances with up to 22,000 edges .
m. goerigk and a. schbel .
algorithm engineering in robust optimization . in l.
kliemann and p. sanders , editors , _ algorithm engineering : selected results and surveys _ , volume 9220 of _ lncs state of the art_. springer , 2016 .
final volume for dfg priority program 1307 .
a. kasperski and p. zieliski .
robust discrete optimization under discrete and interval uncertainty : a survey . in _ robustness analysis in decision aiding , optimization , and analytics _ , pages 113143 .
springer , 2016 . | in classic robust optimization , it is assumed that a set of possible parameter realizations , the uncertainty set , is modeled in a previous step and part of the input . as recent work has shown , finding the most suitable uncertainty set is in itself already a difficult task .
we consider robust problems where the uncertainty set is not completely defined .
only the shape is known , but not its size .
such a setting is known as variable - sized uncertainty . in this work we present an approach how to find a single robust solution , that performs well on average over all possible
uncertainty set sizes .
we demonstrate that this approach can be solved efficiently for min - max robust optimization , but is more involved in the case of min - max regret , where positive and negative complexity results for the selection problem , the minimum spanning tree problem , and the shortest path problem are provided .
we introduce an iterative solution procedure , and evaluate its performance in an experimental comparison .
* keywords : * robust combinatorial optimization ; min - max regret ; variable - sized uncertainty |
rr lyrae ( rrl ) stars have played an important role in determining the cosmic distance scale in modern astronomy . while several studies have used classical cepheids as primary distance indicators ,
rrl stars have received substantial attention in solving the distance scale problem . apart from distance determinations , rrl stars are of particular importance as a test bed for the theories of stellar and galactic structure and evolution .
they are the tracers of old stellar populations of the bulge , disk and halo components that are present everywhere .
rrl stars are radially pulsating a - f variable stars with periods in the 0.2 - 1.2 day range , and amplitude of variation @xmath10 mag .
they can be easily identified , and play a key role as the cornerstone of the population ii distance scale .
they are extensively used to determine distances to old and sufficiently metal - poor systems , where they are commonly found in large numbers . in particular , rrl stars are present in globular clusters ( gcs ) and the dwarf galaxies in the neighborhood of the milky way ( greco et al .
2007 ) , and have also been identified in the m31 field ( brown et al .
2004 , dolphin et al .
2004 ) , in some m31 companions ( pritzl et al . 2005 ) , and in at least four m31 gcs ( clementini et al . 2001 ) .
distances to the large magellanic cloud ( lmc ) for the population ii objects are based on the luminosity of the rrl stars ( clementini et al .
2002 ) .
there have been a very few studies to estimate the distance scale of the smc using the rrl stars .
using four rrl light curves of the smc cluster ngc 121 , walker & mack ( 1988 ) have estimated the distance modulus of the cluster to be [email protected] mag .
on the other hand , using 22 rr lyrae stars surveyed around 1.3 square degrees near the northeast arm of the smc field ngc 361 , smith et al .
( 1992 ) obtained the distance modulus of the smc as [email protected] mag . also , there are other studies of estimating the distance scale of the smc based on the binary star light curves and double mode cepheids .
harries et al .
( 2003 ) reported that the distance modulus to the smc is of the order of [email protected] mag taking 10 eclipsing binaries in the smc . by selecting 40 eclipsing binaries of spectral type o and b in the smc , hilditch et al .
( 2005 ) have derived the fundamental parameters of the binaries and refined the distance modulus to the smc to [email protected] mag .
also kovcs et al .
( 2000 ) found the distance modulus of the smc to be [email protected] mag based on the photometric data of double - mode cepheids from the ogle project .
the stellar atmospheric parameters of effective temperature ( t@xmath11 ) and surface gravity ( log g ) are of fundamental astrophysical importance .
they are the prerequisites to any detailed abundance analysis and define the physical conditions in the stellar atmosphere and hence are directly related to the physical properties mass ( m ) , radius ( r ) and luminosity ( l ) of the star . in this paper , we present an independent analysis of 536 smc rrl stars discovered by the ogle project ( soszysky et al .
the ogle database is a very wealthy resource for studying the characteristics of variable stars in the galaxy , lmc and smc .
for the first time , we make use of the ogle smc data to estimate the distance scale of the smc using a large number of well - sampled rrl light curves .
the road map of the present investigation is to perform a fourier analysis of the rrl stars in order to estimate their physical parameters and hence the smc distance scale .
we employ the fourier decomposition technique which is used extensively to characterize the observed photometric light curves of rrl and other types of variables .
the accurate determination of the fourier coefficients is , therefore , an important task .
we have performed an independent automated fourier analysis of all the rrl light curves selected in this paper by a computer code developed by us . in section 2
we give a brief description of the database that we use and the procedure of removing the outliers from the light curve data .
we present fourier decomposition of the light curves in section 3 .
we also describe the use of the unit - lag auto - correlation function for finding out the optimal order of the fit to the rrl light curves .
section 4 describes the error analysis of the fourier decomposition parameters @xmath12 and r@xmath13 .
section 5 describes the calibration of the i band data to the v band . in section 6
, we describe the various physical parameters of the rrls obtained by using empirical relations from the literature .
section 7 describes the distance determination of the smc .
lastly , in section 8 , we present the conclusions of our study .
the rrl stars analyzed in the present work were selected from the high quality photometric catalog of rrl stars in the smc from ogle - ii database ( soszy ' nski et al .
2002 , hereafter sz02 ) .
the catalog contains b , v and i light curves of 58 rrc stars and 478 rrab stars located in the 11 areas close to the bar of the smc in johnson - cousins photometric system .
the target stars selected for the present analysis were passed through a multi - pass nonlinear fitting algorithm in idl ( interactive data language ) which is very efficient in removing the outliers from the data set .
points lying @xmath14 away from the fit are rejected from the data set so that the objects with well sampled and accurate light curves are available for the analysis .
all the selected targets have evenly covered i band light curves with about 100 - 400 data points in i band with an internal accuracy of 0.03 - 0.13 mag in i band .
the b and v light curves in the sz02 catalog are of lower quality and have a limited number of data points ( 20 - 40 ) .
therefore , we use only the i band data for estimation of the fourier parameters . in fig .
1(a ) and ( c ) we show the typical phased light curve of a rrab and a rrc star in the database .
the corresponding light curves with outliers @xmath15 removed are shown in panels ( b ) and ( d ) respectively .
of 536 rrl stars selected from the database.,width=302,height=226 ] fourier decomposition technique is a powerful and robust technique for describing the shape of the photometric light curves of the rrl and other variable stars .
the method has also been utilized for variable star classification ( deb & singh 2009 ) , finding the physical parameters of stars like absolute magnitude , metallicity , effective temperature , gravity , helium abundance etc .
of rrl and other stars ( kovcs & kupi 2007 and references therein ) .
the method in its modern form has been revived by simon & lee ( 1981 ) to describe the progression of cepheid light curve with increasing period called hertzsprung progression .
they showed that the lower order fourier parameters are able to completely describe the progression of cepheid light curves . with theoretical light curves based on hydrodynamical models
simon & lee ( 1981 ) , simon & clement ( 1993 , hereafter sc93 ) , clement & rowe ( 2000 ) were able to derive various physical parameters of stars . on the other hand kovcs and his collaborators ( jurcsik & kovcs 1996 , hereafter jk96 , kovcs & jurcsik 1996 , jurcsik 1998 , kovcs 1998 , kovcs & kanbur 1998 , kovcs & walker 1999 ) were able to estimate astrophysical parameters by establishing empirical relationships between the fourier parameters and the physical parameters of stars in the field .
a number of studies have made use of these empirical relations to derive the physical parameters of the rrl stars .
to list a few of these studies , kaluzny et al .
( 2000 ) used these empirical relations to estimate physical parameters of 26 rrab and 16 rrc stars in m5 , pea et al .
( 2007 ) used these empirical relations to estimate physical parameters of 7 rrl stars in bootes and arellano et al .
( 2008 ) have determined parameters for few rrl variables in ngc 5466 .
there are a number of other studies that have made use of the fourier coefficients to determine various parameters of astrophysical importance .
sandage ( 2004 ) studied the correlation of particular fourier components of the light curves of rrab stars with metallicty discovered by simon and later by kovcs and his co - workers . on the other hand , morgan et al .
( 2006 , 2007 ) have derived empirical relations for rrc stars connecting @xmath16 $ ] , period ( p ) and fourier parameter @xmath17 . in all the above studies , it has been seen that fourier parameters can indeed be linked with global stellar parameters such as luminosity , mass , temperature , metallicty and radius .
most of the physical parameter estimations of stars require accurate calculation of fourier parameters and any inaccuracies in the determination of fourier parameters will reflect in the calculated physical parameters of the stars .
we computed fourier decomposition coefficients for 536 rrl stars in the smc for the i band photometric data .
the observed magnitudes were fitted with a fourier cosine series of the form @xmath18 where @xmath19 is the observed magnitude , @xmath20 is the mean magnitude , @xmath21=2@xmath22 is the angular frequency , @xmath23 is the period of the star in days , @xmath24 is the time of observation , @xmath25 is the epoch of maximum light , @xmath26 and @xmath27 are the @xmath28th order fourier coefficients and @xmath29 is the order of the fit .
( 1 ) has @xmath30 unknown parameters which require at least the same number of data points to solve for these parameters .
since period is known from the database , the observation time can be folded into phase ( @xmath31 ) as @xmath32 here @xmath25 is the epoch of maximum light of the rrl light curves .
the value of @xmath31 is from 0 to 1 , corresponding to a full cycle of pulsation and and @xmath33 denotes the integer part of the quantity .
hence , eqn .
( 1 ) can be written as ( schaltenbrand & tammann 1971 ) @xmath34.\ ] ] the fourier parameters are defined as @xmath35 where @xmath36 .
the @xmath37 values have been adjusted to lie between 0 and 2@xmath38 so that they are comparable to those available in the literature .
the optimal order of the fit ( @xmath39 ) was chosen to be 4 for rrc stars and 5 for rrab stars by the calculation of unit - lag auto - correlation function ( fig . 2 ) and looking at the @xmath40 value of the fit .
a detailed discussion on the optimal order of the fit has been given in deb & singh ( 2009 ) which involves the calculation of unit - lag auto - correlation function using baart s condition ( baart 1982 , petersen 1986 ) .
increasing the order of the fit may reduce the @xmath41 to some extent but this will underestimate the distribution .
the fits to the phase - folded light curves were made using levenberg - marquardt ( lm ) algorithm which is based on @xmath41 minimization method ( press et al .
the reduced chi - square for the fit to the rrl light curves are nearly 1 .
the histogram plot of the @xmath42 of the rrl stars is shown in fig . 3 , where @xmath43 is the degree of freedom of the fit which is defined as the number of data points minus the number of parameters used to fit the light curve data .
complete table is available in the electronic form .
complete table is available in the electronic form . in figs .
4 & 5 we show smoothed ( outliers removed ) fourier fitted light curves for 16 randomly selected rrab and rrc stars respectively .
further , the fourier parameters for all the stars computed from the i band data are given in table 1 ( for rrab ) and table 2 ( for rrc ) .
@xmath44 denotes the number of data points for each of the light curves retained for the analysis after removing the outliers .
@xmath45 is the deviation parameter for rrab stars and is discussed in sec .
( 6.1 ) . in fig .
6 we plot the fourier amplitude ratios @xmath46 , @xmath47 , @xmath48 and phase differences @xmath49 , @xmath50 , @xmath51 versus @xmath52 for the rrl dataset . and
@xmath17 as a function of log p for the i band data of 536 program stars .
open circles represent rrab stars and filled circles rrc stars.,width=302,height=302 ]
the error in @xmath53 , where the standard errors in @xmath54 and @xmath55 are known from lm method , is given by @xmath56 in the case of a large number of observations eqn .
( 3 ) can be reasonably approximated by @xmath57 therefore the errors in r@xmath13 and @xmath37 can be approximately written as @xmath58 @xmath59 in figs .
7 & 8 we plot the distribution of estimated errors in the quantities r@xmath60 , r@xmath61 , r@xmath62 and @xmath63 , @xmath17 and @xmath64 . as is quite obvious that the errors in @xmath17 are larger than the errors in @xmath63 because the amplitudes for the higher order coefficients are smaller and hence it is very difficult to derive their phase with better precision . ,
r@xmath61 and r@xmath62 for the i band data .
, width=302,height=302 ] , @xmath17 and @xmath64for the i band data .
, width=302,height=302 ]
the fourier coefficients of a particular light curve in different photometric bands are different .
nonlinear hydrodynamical models have been used by dorfi & feuchtinger ( 1999 , hereafter df99 ) to obtain ubvi light curves of rrl stars .
they have computed the inter - relations of the fourier parameters in v and i bands and compared them with those obtained by morgan et al .
( 1998 , hereafter msb98 ) from observed light curves of metal poor globular cluster m68 by walker ( 1994 , hereafter w94 ) .
however , only 8 rrab and 8 rrc stars from w94 were used for setting up the relations in msb98 . while the v - i inter - relations for amplitude ratios ( @xmath65 ) of df99 showed good agreement with the empirical relations of msb98 , large deviations showed up in the relations of some of the phase differences ( @xmath37 ) for two sets of models with z = 0.001 and 0.0001 .
df99 concluded that no distinct metallicity effect could be followed from these results and emphasized the need for more model calculations covering a wide range of stellar parameters . in order to check the reliability of the theoretical inter - relations of the phase parameters of df99
, we have calculated the fourier phase parameters in the i and v bands from the highly accurate photometric light curves of selected rrab and rrc stars in globular cluster m3 from benk et al .
( 2006 , hereafter b06 ) together with the data of 8 rrab and 8 rrc stars from w94 .
all the rrab stars free from blazhko effect and rrc stars having good quality light curves were selected from b06 after visual inspection .
we have used 29 rrab variables 1 , 6 , 9 , 25 , 27 , 31 , 32 , 42 , 53 , 57 , 58 , 69 , 84 , 89 , 93 , 94 , 100,109,135,137 , 139 , 142 , 144 , 146 , 148 , 165 , 167 , 175 , 222 of globular cluster m3 from b06 and 8 variables 2 , 9 , 10 , 14 , 22 , 23 , 25 , 35 of globular cluster m68 from w94 . in the case of rrc stars , 16 variables 21 , 37 , 75 , 86 , 88 , 107 , 126 , 128 , 131 , 147 , 152 , 171 , 177 , 208 , 213 , 259 of globular cluster m3 from b06 and 8 variables 1 , 6 , 11 , 13 , 16 , 18 , 20 , 24 of globular cluster m68 from w94 were used .
we call this data set as b06+w94 . in fig
. 10 , we show the v - i inter - relations of the fourier phase parameters @xmath66 for rrab and rrc stars respectively .
points lying above a threshold sigma ( @xmath67 ) from the fit are rejected using multi - pass fitting algorithm in idl .
the threshold sigma level is chosen based on a reasonably high degree of correlation between the parameters in the i and v band so that the number of data points for the fit retained are significant .
the mean difference of 0.61% in absolute magnitude , 0.23% in @xmath68 , 4.03% in [ fe / h ] , 0.10% in effective temperature and 1.18% in mass will result when df99 inter - relations are used .
the fourier inter - relations of the phase parameters from the i band to the v band have the following form : @xmath69 in tables 3 and 4 , we list the inter - relations obtained from the b06+w94 dataset for rrab and rrc stars . also given are the theoretical inter - relations of df99 . in fig .
10 , we plot the i and v band fourier parameters . in fig . 11 , we plot the theoretical inter - relations between the i and v band of df99 and observational inter - relations obtained from the b06+w94 dataset . to compute v band amplitude from i band amplitude we again resort to the existing work in the literature .
12 we plot v band amplitudes as functions of i band amplitudes for rrl stars in 4 globular clusters , m68 from walker ( 1994 ) , ic 4499 from walker & nemec ( 1996 ) , ngc 1851 from walker ( 1998 ) and m3 from benk et al .
solid line denotes the linear relation as determined by a least square fit given by @xmath70 calibration to the standard v band magnitude is necessary for calculation of the distance scale .
the mean magnitude was calculated for each object from its light curve . to transform the i band mean magnitude to the v band we used rrl light curves in the v and i band from the ogle database itself .
typical errors in the v band magnitudes in ogle are 0.02 - 0.07 mag . for rrab stars
a linear least - square fit resulted in the following conversion relation : @xmath71 for the rrc stars the result of the linear least square fit is as follows : @xmath72 the fits of eqns .
( 9 ) & ( 10 ) are shown in fig .
light curves having mean magnitude variation @xmath73 were discarded and v band magnitude thus determined .
on the other hand , the absolute magnitude calculation of rrab stars required a@xmath74 while rrc stars required a@xmath75 in v band .
we used v and i band data from walker ( 1994 , 1998 ) for the a@xmath75 v band calibration and a linear regression analysis yielded the following relation @xmath76 for rrab stars and @xmath77 for rrc stars . following saha & hoessel ( 1990 ) and sakai et al .
( 1999 ) , the intensity weighted mean magnitudes and phase - weighted mean magnitudes are given by @xmath78 @xmath79 + where n is the total number of observations , m@xmath80 and @xmath81 are the magnitude and phase of the @xmath28th observation respectively , in order of increasing phase .
the intensity - weighted mean magnitude and phase - weighted mean magnitude in the v band have been determined using a linear least square fit between the v and i band data resulting in the following conversion relations @xmath82
@xmath83 for rrab stars , and @xmath84 @xmath85 for rrc stars .
using a set of 81 rrab stars , jk96 derived an empirical relation between the fourier decomposition parameter @xmath86 , period ( p ) and metallicity [ fe / h ] given by @xmath87 = -5.038 - 5.394\ , \rm p + 1.345\ , \rm \phi_{31}.\ ] ] the fourier parameters in the above relation were obtained for a sine fourier series fit .
care has to be taken that the rrab stars to which this equation is applied follow the light curve systematics defined by calibrating the sample of jk96 .
this can be ensured if the light curve in question satisfies a certain compatibility criterion .
the various parameters of the calibrating data set obey well - defined correlations ( see table 6 of jk96 ) .
deviation of a light curve from the calibrating characteristics is quantified via the deviation parameter d@xmath88 defined as @xmath89 where f@xmath90 is the observed value of a given fourier parameter , @xmath91 is the corresponding deviations of various correlations .
if the maximum value of d@xmath92 of each parameter is less than 3 , then each light curve satisfies the compatibility condition .
this maximum value of d@xmath93 represents a quality test on the regular behaviour of the shape of the light curve .
but when the test is applied to the rrab stars of ogle smc rr lyrae stars , some of the stars did not pass this criterion even though the shapes of the light curve are ` normal looking ' .
one of the possible reasons is that the ogle smc rrl light curves are constructed from the observation in the i band but not in the v band on which the deviation parameter is defined .
we have noticed that for the deviation parameter up to a value of d@xmath94 , the shapes of the rrab light curves are normal .
this value of d@xmath93 is used as a cutoff limit for the selection of clean samples of rrab stars for analysis .
the deviation parameter of individual rrab light curves of the ogle data has been calculated and is given in the last column of table 1 .
cacciari et al .
( 2005 ) have also adopted this slightly relaxed criterion to improve the statistics after they verified that this does not lead to any significant difference in the resulting physical parameters .
for the rrab analysis , we apply various empirical relations from the literature to all our target stars only when d@xmath95 . after implementing this compatibility test
we have found that 355 rrab stars out of 478 are now available for further analysis .
this means that our sample of rrab stars now contains 355 rrab stars that have ` normal looking ' light curve shapes .
it is also seen that the maximum contribution to the deviation parameter comes from the deviation in @xmath17 .
the intrinsic colors as derived from the fourier parameters of the rrab stars can be used to estimate the temperature of these stars .
these color indices as defined by jurcsik ( 1998 ) are the differences of the magnitude - averaged absolute brightness .
the intrinsic color relations is of the form @xmath96 the above relation can be used to calculate the effective temperature @xmath97.\end{aligned}\ ] ] the relation between the absolute magnitudes and the fourier parameters is given by the following relations @xmath98 eqn .
( 20 ) is from jurcsik ( 1998 ) and eqn .
( 21 ) is taken from kovcs & walker ( 2001 ) .
( 22 ) , taken from kovcs ( 1998 ) is based on the relation between the intensity - averaged m@xmath99 and the v band fourier parameters . on the other hand on a different absolute scale , jurcsik ( 1998 )
derived a relation for the mass of the rrab stars as follows @xmath100.\end{aligned}\ ] ] the fourier coefficients in the above equations are based on a sine series fit .
but the coefficients in the present analysis ( table 1 ) are calculated for a cosine series fit . therefore , to put the coefficient @xmath50 on to the appropriate system , we add 3.145 to @xmath50 in table 1 . in table 5
we list various physical parameters computed from the above empirical relations for 335 rrab stars with regular light curves with @xmath101 .
mean error in the fourier phase parameter @xmath17 for 335 rrab stars is @xmath102 0.19 .
this should result in an uncertainty of @xmath102 0.25 dex in [ fe / h ] , @xmath102 0.02 mag in the absolute magnitude , @xmath102 12 k in temperature , @xmath102 0.01 m@xmath2 in mass .
for the rrc stars , we have only retained stars for which the error in @xmath17 is less than 0.5 .
it is generally not desirable to have error larger than 0.2 ( kaluzny et al . 1998 ) .
since the rrc stars in the smc are fainter and have lower amplitudes , good data for such sources are simply not available at present . imposing the criterion of @xmath103 leaves us with only 3 rrc stars to be used for further analysis . to increase the statistics without significant loss of accuracy
, we have chosen @xmath104 which allows us to analyse 17 rrc stars . through hydrodynamical pulsation models
, sc93 formulated the following relationships for luminosity , mass and helium abundance ( y ) of rrc stars : @xmath105 @xmath106 @xmath107 @xmath108 the rms error to the fit in eqn .
( 24 ) is 0.025 .
another alternative to eqn .
( 24 ) is the expression for the intensity averaged absolute magnitude proposed by kovcs ( 1998 , hereafter k98 ) based on observations of rrc variables in eight different stellar systems : @xmath109 with an rms of 0.042 .
kovcs derived this formula from the light curves of 106 rrc stars , mainly from sculptor and m68 and calibrated the zero point from the baade - wesselink luminosity scale of clementini et al .
the phase difference @xmath63 in eqn .
( 28 ) is based on a sine series fit .
therefore , 1.571 is subtracted from @xmath63 values in table 2 , which are based on a cosine series fit .
following lzaro et al .
( 2006 ) , we have calculated the luminosities of smc rrc stars from eqn .
( 24 ) as well as from eqn .
the results are plotted in fig .
luminosities calculated using sc93 relation ( eqn . ( 24 ) ) are larger than those estimated by the k98 relation ( eqn .
( 28 ) ) and the difference may be attributed to different scaling ( zero points ) of the two relations .
the zero point of the luminosity scale of rrc stars is still a matter of controversy ( cacciari et al . 2005 , nemec 2004 , lzaro et al . 2006 ) .
some problems with the simon s calibration for the rrc stars were also noted by catelan ( 2004 ) who concluded that sc93 s relations for l and m can not both be simultaneously valid . eqns . of sc93 ( eqns .
( 24 ) & ( 25 ) ) yield
@xmath110 the above relation lacks a temperature term which should be present according to the period - density relation ( catelan 2004 ) .
this is true and remains an unresolved theoretical issue .
the empirical relation for metallicity in terms of the fourier decomposition parameter @xmath50 was derived by morgan et al .
( 2007 ) by considering a large number of stars in different globular clusters which have the v band fourier coefficients available in the literature and the sources of metallicities available for these cluster members . using the metallicty scales of zinn & west ( 1984 ) and carretta & gratton ( 1997 )
, they obtained the following two empirical relations for rrc stars respectively which has a sample standard deviation of 0.145 dex , and @xmath87 = 0.034\,\rm { \phi_{31}}^2 + 0.196\,\rm \phi_{31 } -8.507\,p + 0.367,\ ] ] which has a sample standard deviation of 0.142 dex .
we have used eqn .
( 31 ) for the calculation of [ fe / h ] . the temperatures determined using eqn .
( 26 ) are found to be overestimated for some of the rrc stars .
cacciari et al .
( 2005 ) also used eqn .
( 26 ) and found a similar trend for temperature estimates of rrc stars of globular cluster m3 when compared with temperatures derived by sekiguchi & fukugita ( 2000 ) from the b - v colors .
temperature differences as large as 500 k have been found for some of the rrc stars .
therefore , the temperature of rrc stars as determined from fourier coefficients need to be re - examined .
the inaccuracy of the empirical relations connecting the effective temperature , @xmath50 term and the period is also reflected in the calculation of the radius of rrc stars .
we show ahead that the radii of rrc stars determined from the fourier decomposition technique differ largely from those determined from the theoretical period - radius - metallicity relation of marconi et al .
on the other hand , for rrab stars there is a well - matched similarity between the radii determined from the fourier decomposition technique and from the more fundamental relations used by marconi et al .
therefore , the empirical relations derived for the rrab stars seem to be accurate , whereas the physical parameter estimation using the empirical relations derived for the rrc stars may be unreliable . in table 6 , we list the various physical parameters computed from the above empirical relations for the rrc stars with @xmath104 .
mean error in the fourier phase parameter @xmath17 for 17 rrc stars is @xmath102 0.396 .
this will result in an uncertainty of @xmath102 0.14 dex in [ fe / h ] , @xmath102 @xmath112 in @xmath113 , @xmath102 16 k in temperature and @xmath102 0.18 m@xmath2 in mass .
an h - r diagram of the smc rrls stars is plotted in fig .
15 along with the theoretical blue and red edges of the instability strip .
the rrc stars are represented by solid circles whereas the rrab stars are represented by open circles .
also shown are the blue and red edges of the instability strip of fundamental ( rrab ) and first - overtone ( rrc ) variables derived from the theoretical convective pulsation models of bono , caputo & marconi ( 1995 ) .
it can be easily seen that all the rrab and rrc stars lie well within their instability strips .
although the rrc stars lie inside the instability strip , the empirical relations overestimate the temperatures of these stars .
calculated from eqn .
( 28 ) and filled upper triangles rrc stars with @xmath68 calculated from eqn .
the green solid and dashed lines show the boundaries of the instability strip for rrc and rrab stars respectively for mass 0.75m@xmath2 and the black solid and dashed lines show the boundaries of the instability strip for rrc and rrab stars respectively for mass 0.65m@xmath2.,width=321,height=302 ] from the global physical parameters like mass , luminosity and temperature , the gravity can be calculated from the following equation ( cf .
cacciari et al . 2005 ) @xmath114 ) ( mar)=-0.112(@xmath115)+1.206(@xmath116)log ( r / r@xmath2)(fd ) and for the rrc stars ( filled circles ) in the lower panel the linear regression analysis yields : log ( r / r@xmath2)(mar)= 0.235(@xmath117)+0.651(@xmath118)log ( r / r@xmath2 ) ( fd).,width=302,height=377 ] the radii of the rrl can be obtained once we know the temperature and the luminosity of the rrls stars using the stefan - boltzmann law .
marconi et al .
( 2005 ) have recently given a new theoretical period - radius - metallicity relation for the rr lyraes based on detailed and homogeneous set of nonlinear models with a wide range of stellar masses and chemical compositions .
we show how the radii obtained by the fourier decomposition method compare to the radii obtained from the period - radius - luminosity ( prz ) relation of marconi et al .
( 2005 ) . for the sake of completeness
, their prz relations are given as follows : @xmath119 for rrab stars with @xmath120 .
@xmath121 for rrc stars with @xmath122 .
@xmath123 can be calculated from the relation @xmath124 - 1.70 + \log \ , ( 0.638\,\rm f + 0.362),\end{aligned}\ ] ] where @xmath125 is an @xmath126-enhancement factor with respect to iron ( salaris , chieffi & straniero 1993 ) .
we take @xmath127 . in fig .
16 , we compare our determinations of the radii using the fourier decomposition technique ( fd ) with those obtained by marconi et al .
( 2005 ) empirical relations ( mar ) . in case of the radius determinations of rrc stars by the fd method ,
the sc93 s equations for @xmath128 and @xmath129 are used .
it can be seen that the radii obtained by the two different methods are quite consistent for rrab stars . the solid line in fig .
16 represents the best least square fit .
the correlation coefficient between the two radii determinations for the rrab stars are 0.994 , whereas for the rrc stars ( lower panel of fig .
16 ) , the correlation coefficient is 0.880 .
this suggests that the radii determinations by the two independent methods are strongly correlated for the rrab stars , whereas this is not the case for the rrc stars .
we use mean magnitude a@xmath130(v ) , intensity - weighted mean magnitude and phase weighted mean magnitude separately to derive an independent distance modulus to the smc .
the distance modulus for each star is calculated from their values of mean magnitudes and m@xmath131 derived from the fourier parameters .
assuming the reddening estimates of the eleven smc fields of sz02 to be accurate , we take average of the reddening values of these fields as the true estimates of e(b - v ) .
following sz02 we take the interstellar extinction as a@xmath131=3.24e(b - v ) .
this value of a@xmath131 has been used to estimate the reddening free distance modulus of smc . using 335 rrab and 17 rrc stars we find the mean distance modulii of smc to be [email protected] mag , 18.87@xmath6 0.03 mag and 18.87 @xmath132 mag from the mean magnitude , intensity - weighted mean magnitude and phase - weighted mean magnitude respectively for all the 352 rrl stars .
the uncertainty is the standard deviation of the mean from the individual stars .
the mean distance modulus so determined are consistent with the earlier values of 19.05@xmath6 0.017 ( kovcs et al .
2000 ) , [email protected] ( harries et al . 2003 ) and [email protected] ( hilditch et al . 2005 ) .
in this paper we have derived the physical parameters of 352 rr lyrae stars ( 335 rrab and 17 rrc ) of the smc from ogle - ii i band database using the fourier decomposition of their light curves .
the stars were selected based on the quality of their light curves .
i band fourier coefficients have been converted to v band using the inter - relations obtained from the observational data of b06 and w94 . using the fourier decomposition method
, we find the mean physical parameters : [ fe / h ] = -1.56@xmath0 , m = 0.55 @xmath10.01m@xmath2 , t@xmath3 k , @xmath4 and m@xmath5 for 335 rrab variables and [ fe / h ] = -1.90@xmath6 0.13 , m = 0.82 @xmath10.18m@xmath2 , t@xmath7 k , @xmath8 and m@xmath9 for 17 rrc stars .
mean distance modulus to the smc was calculated from the 352 light curves by using mean magnitude , intensity weighted mean magnitude and phase weighted mean magnitude .
the values of the distance modulus are found to be in good agreement with independent studies .
locations of the rrl stars in the h - r diagram clearly show that the estimates of the parameters determined from the fourier decomposition method are consistent with the theoretical blue and red edges of the instability strip calculated by bono et al .
the calculations of the radii of the rrls by using the fourier decomposition technique are in agreement with the theoretical period - radius - metallicity relations of marconi et al .
( 2005 ) obtained from a completely different approach of non - linear convective modelling .
complete table is available in the electronic form .
sd thanks csir , govt . of india for a senior research fellowship .
hps is grateful to cral - observatoire de lyon for an invited professorship .
authors thank the reviewer dr .
horace smith for many useful comments and suggestions which improved the presentation of the paper .
arellano ferro , a. , rojas lpez , v. , giridhar , s. , bramich , d. m. 2008 , mnras , 384 , 1444 baart , m. l. , 1982 , i m a j. num .
analysis , 2 , 241 benk , j. m. , bakos , g. , nuspl , j. 2006 , mnras , 372 , 1657 ( b06 ) bono , g. , caputo , f. , marconi , m. 1995 , aj , 110 , 2365 brown , t. m. , ferguson , h. , smith , e. , kimble , r. a. , sweigart , a. v. , renzini , a , rich , r. m. 2004 , aj , 127 , 2738 cacciari , c. , corwin , t. m. , carney , b. e. , 2005 , aj , 129,267 carretta , e. , gratton , r. g. , 1997 , a&as , 121 , 95 catelan , m. , 2004 , aspc , 310 , 113 clement , c. m. , rowe , j. , 2000 , aj , 120 , 2579 clementini , g. , carretta , e. , gratton , r. , merighi , r. , mould , j. , r. , & mccarthy , j. , k. , 1995 , aj , 110 , 2319 clementini , g. , federici , l. , corsi , c. , cacciari , c. , bellazzini , m. , smith , h. a. , 2001 , apj , 559 , l109 clementini , g. , bragaglia , a. , di fabrizio , l. , maio , m. et al .
, 2002 , aspc , 259 , 124 deb , s. , singh , h. p. , 2009
, a&a ( in press ) ; arxiv:0903.3500v3 dorfi , e. a. , feuchtinger , m. u. , 1999 , a&a , 348 , 815 ( df99 ) dolphin , a. e. , saha , a. , olszewski , e. w. , thim , f. , skillman , e. d. , gallagher , j. s. , hoessel , j. , 2004 , aj , 127 , 875 greco , c. , clementini , g. , catelan , mrcio , held , e. , v. et al . , 2007 , apj , 670 , 332 harries , t. j. , hilditch , r. w. , howrath , i. d. , 2003 , mnras , 339 , 157 hilditch , r. w. , howrath , i. d , harries , t. j. , 2005 , mnras , 357 , 304 jurcsik , j. , kovcs , g. , 1996 , a&a , 312 , 111 ( jk96 ) jurcsik , j. , 1998 , a&a , 333 , 571 kaluzny , j. , hilditch , r. w. , clement , c. , rucinski , s. m. , 1998 , mnras , 296 , 347 kaluzny , j. , olech , a. , thompson , i. , pych , w. , krzeminski , w. , schwarzenberg - czerny , a. , 2000 , a&as , 143 , 215 kovcs , g. , jurcsik , j. , 1996 , apj , 466 , l17 kovcs , g. , jurcsik , j. , 1997 , a&a , 322 , 218 kovcs , g. , 1998 , mem .
soc . astron .
italiana , 69 , 49 ( k98 ) kovcs , g. , 2000 , a&a , 360 , l1 kovcs , g. , walker , a. r. , 1999 , apj , 512 , 271 kovcs , g. , walker , a. r. , 2001 , a&a , 374 , 264 kovcs , g. 2005 , a&a , 438 , 227 kovcs , g. , kanbur , s. m. 1998 , mnras , 295 , 834 kovcs , g. , kupi , g. 2007 , a&a , 462 , 1007 lzaro , c. , arellano ferro , a. , arvalo , m. j. et al .
2006 , mnras , 372 , 69 marconi , m. , nordgren , t. , bono , g. et al .
2005 , apj , 623 , l133 morgan , s. m. , simet , m. , bargenquast , 1998 , acta astron . , 48 , 341 ( msb98 ) morgan , s. m. , wahl , j. n. , wieckhorst , r. m. , 2006 , mmsai , 77 , 178 morgan , s. m. , wahl , j. n. , wieckhorst , r. m. , 2007 , mnras , 374 , 1421 nemec , j. , m. , 2004 , aj , 127 , 2185 pea , j. h. , arellano , a. , sareyan , j. p. , pea , r. , alvarez , m. 2007 , comm .
in asteroseismology , 150 , 385 petersen , j. o. , 1986 , a&a , 170 , 59 press , w. , teukolsky , s. , vetterling , w. , flannery , b. 1992 , numerical recipes in fortran ( 2nd ed . ; cambridge : cambridge univ .
press ) pritzl , b. j. , armandroff , t. e. , jacoby , g. h. , da costa , g. s. , 2005 , aj , 129 , 2232 saha , a. , hoessel , j. g. , 1990 , aj , 99 , 97 sakai , s. , ferrarese , l. , kennicutt , r. c. et al .
, 1999 , apj , 523 , 540 salaris , m. , chieffi , a. , straniero , o. , 1993 , apj , 350 , 645 sandage , a. , 2004 , aj , 128 , 858 schaltenbrand , r. , tammann , g. , 1971 , a&as , 4 , 265 sekiguchi , m. , fukugita , m. , 2000 , aj , 120 , 1072 simon , n. r. , clement , c. m. , 1993 , apj , 410 , 526 ( sc93 ) simon , n. , lee , a. 1981 , apj , 248 , 291 smith , h. a. , silbermann , n. a. , baird , s. , r. , graham , j. a. , 1992 , aj , 104 , 1430 soszy ' nski , i. , udalski , a. , szymaski , m. et al .
, 2002 , acta astron .
, 52 , 369 ( sz02 ) walker a. r. , 1994 , aj , 108 , 555 ( w94 ) walker a. r. , nemec , j. m. , 1996 , aj , 112 , 2026 walker a. r. , 1998 , aj , 116 , 220 walker , a. r. , mack , p. , 1988
, aj , 96 , 872 zinn , r. , west , m. j. , 1984 , apjs , 55 , 45 | we present a careful and detailed light curve analysis of rr lyrae stars in the small magellanic cloud ( smc ) discovered by the optical gravitational lensing experiment ( ogle ) project . out of 536 single mode rr lyrae stars selected from the database
, we have investigated the physical properties of 335 ` normal looking ' rrab stars and 17 rrc stars that have good quality photometric light curves .
we have also been able to estimate the distance modulus of the cloud which is in good agreement with those determined from other independent methods .
the fourier decomposition method has been used to study the basic properties of these variables .
accurate fourier decomposition parameters of 536 rr lyrae stars in the ogle - ii database are computed .
empirical relations between the fourier parameters and some physical parameters of these variables have been used to estimate the physical parameters for the stars from the fourier analysis .
further , the fourier decomposition of the light curves of the smc rr lyrae stars yields their mean physical parameters as : [ fe / h ] = -1.56@xmath0 , m = 0.55 @xmath10.01m@xmath2 , t@xmath3 k , @xmath4 and m@xmath5 for 335 rrab variables and [ fe / h ] = -1.90@xmath6 0.13 , m = 0.82 @xmath10.18m@xmath2 , t@xmath7 k , @xmath8 and m@xmath9 for 17 rrc stars . using the absolute magnitude together with the mean magnitude , intensity - weighted mean magnitude and the phase - weighted mean magnitude of the rr lyrae stars ,
the mean distance modulus to the smc is estimated to be [email protected] mag , [email protected] mag and [email protected] mag respectively from the rrab stars . from the rrc stars ,
the corresponding distance modulus values are found to be [email protected] mag , [email protected] mag and [email protected] mag respectively .
since fourier analysis is a very powerful tool for the study of the physical properties of the rr lyrae stars , we emphasize the importance of exploring the reliability of the calculation of fourier parameters together with the uncertainty estimates keeping in view the large collections of photometric light curves that will become available from variable star projects of the future .
[ firstpage ] stars : rr lyrae : methods : data analysis |
[ [ section ] ] five books have so far been published in george r. r. martin s popular _ song of ice and fire _ series @xcite , @xcite , @xcite , @xcite , @xcite .
it is widely anticipated that two more will be published , the first of which ( * ? ? ? * foreword ) will be entitled _ the winds of winter _
each chapter of the existing books is told from the point of view of one of the characters .
so far , ignoring prologue and epilogue chapters in some of the books , @xmath0 characters have had chapters told from their point of view . a chapter told from the point of view of a particular character @xmath1
will be called a _
pov chapter for _ @xmath1 .
a character who has at least one pov chapter in the series will be called a _ pov character . _ [ [ section-1 ] ] the goal is to predict how many pov chapters each of the existing pov characters will have in the remaining two books ( especially @xcite . )
this varies from character to character , since some major characters have been killed off and are unlikely to appear in future novels , whereas other characters are of minor importance and may or may not have chapters told from their point of view .
[ [ section-2 ] ] no attempt is made to deal with characters who have not yet appeared as pov characters .
this issue is discussed further in section [ newcharacters ] .
the data consist of a @xmath2 matrix @xmath3 obtained from http://www.lagardedenuit.com/ , a french fansite .
the rows of @xmath3 correspond to pov characters and the columns to the existing books in order of publication .
the @xmath4entry of @xmath3 is the number @xmath5 of pov chapters for character @xmath6 in book @xmath7 .
the data are displayed in table [ m ] .
.the data .
novel titles are abbreviated using their initials , so for example agot = _ a game of thrones _ etc .
obtained from http://www.lagardedenuit.com/wiki/index.php?title=personnages_pov .
a similar table appears at http://en.wikipedia.org/wiki/a_song_of_ice_and_fire . [ cols="<,^,^,^,^,^",options="header " , ] 2ex it is easy to predict how many pov chapters certain characters will have in the next book . for example , if character @xmath1 was beheaded in book @xmath8 and has not appeared in subsequent books , most readers would predict that @xmath1 will have @xmath9 pov chapters in book @xmath10 .
if we denote by @xmath11 the number of pov chapters for @xmath1 in book @xmath10 , we predict @xmath12 .
the prediction @xmath12 is a _ point prediction .
on the other hand , suppose it is believed that a certain character @xmath13 will have @xmath14 pov chapters in book @xmath10 .
it is quite plausible that @xmath13 might have @xmath15 or @xmath16 pov chapters instead . on the other hand
, it may be thought unlikely that @xmath13 will appear in @xmath9 ( too few ) or @xmath17 ( too many ) pov chapters . instead of the point prediction @xmath18
, it would be better to give a range of likely values for @xmath19 this is an _ interval prediction . _ for example
, to say that the interval @xmath20 $ ] is an @xmath21 credible interval for @xmath22 is to say that there is an @xmath21 probability that @xmath23 .
even more refined than an interval of likely values is a probability distribution over all possible values of the number of pov chapters .
for example , using @xmath24 to denote probability , we could say @xmath25 , @xmath26 , @xmath27 , and @xmath28 .
this probability distribution completely describes our belief about @xmath22 .
conceptually , it can be thought of as a guide for betting .
for example , if our beliefs are correct then it is an even money bet that @xmath18 and also an even money bet that @xmath29 .
our aim is to give such a probability distribution for each pov character @xmath13 .
how can we get a probabilistic prediction as described in section [ probabilisticprediction ] ?
one way would be to assign probabilities based on our gut feelings about the probability of different outcomes , as a traditional bookmaker might do .
( an uncharitable reader might even argue that this would be better than the approach taken below . )
an alternative approach is to choose a statistical _ model _ , that is , a process which could plausibly have generated the observed data .
once the model has been chosen , it can be used to predict future data . in section [ model ] , we describe a family of possible models which depend on six parameters .
values of these parameters which are likely to have produced the observed data are found , and these parameters are used to generate predictions for future books .
this step is repeated many times to build up probability distributions for the predictions .
the whole process of finding values of the parameters is called _ inference _ or _ fitting the model . _ [ [ section-3 ] ] in general ,
the best predictions are obtained by a combination of modelling and common sense . here
we focus entirely on the modelling side and leave common sense behind .
the question to be answered could be expressed as : what could be predicted about future books if we knew nothing about the existing books except for table [ m ] ? "
denote the number of pov chapters for character @xmath6 in book @xmath30 by @xmath31 , @xmath32 .
we assume that there are times @xmath33 and @xmath34 such that the character is ` on - stage ' between @xmath33 and @xmath34 and ` off - stage ' at other times .
for example , the character might be killed off at time @xmath34 . in other words , @xmath35 for @xmath36 and @xmath37 .
for @xmath38 we assume that pov chapters follow a poisson distribution with parameter @xmath39 .
[ [ section-4 ] ] it is inconvenient to have @xmath33 and @xmath34 as model parameters , so instead we assume that there are @xmath40 and @xmath41 such that @xmath42 this is the same as putting @xmath43 and @xmath44 in section [ descriptionofmodel ] .
note that we allow @xmath45 and @xmath41 to take real values , even though @xmath30 is constrained to be an integer .
[ [ section-5 ] ] it is undesirable for the @xmath45 , @xmath41 and @xmath39 for each character to be independent of the other characters as this would give a model with @xmath46 parameters where @xmath47 is the number of characters .
it is unlikely that good predictions could be made from a model with too many parameters . to cut down the number of parameters , we assume that @xmath39 , @xmath41 and @xmath45 are _ random effects , _ which means that they are samples from some underlying probability distribution .
one motivation behind this assumption is that the parameters for different characters are assumed to have something in common .
for example , there might be a typical value for @xmath45 which reflects how long the average character is likely to last in _ a song of ice and fire .
_ [ [ section-6 ] ] the @xmath48 are assumed to be normally distributed and @xmath41 and @xmath45 are also assumed to be normally distributed . however , if there are no constraints on the values of @xmath41 and @xmath45 , the model becomes difficult to fit , because for example there would be no difference in data generated by @xmath49 and @xmath50 for a particular character @xmath6 , regardless of the value @xmath39 . because this makes inference problematic
, we assume that the @xmath51 and @xmath52 distributions are truncated in the interval @xmath53 $ ] . the overall model is : @xmath42 for @xmath54 , and @xmath32 , with @xmath55\\ \beta_i & \sim n(\mu_\beta , \sigma_\beta^2 ) \text { truncated to } [ 0,7]\end{aligned}\ ] ] where @xmath56 and @xmath57 . for fixed @xmath6 , the @xmath31
are assumed to be conditionally independent given @xmath58 and @xmath41 .
for fixed @xmath30 and @xmath59 , the @xmath31 and @xmath60 are assumed to be conditionally independent given the values of @xmath61 and @xmath62 and @xmath63 .
[ [ section-7 ] ] to be explicit , let @xmath64 be the data . for @xmath54 define @xmath65 @xmath66 by @xmath67
then the likelihood is proportional to @xmath68 where the integral is over the @xmath46 dimensions @xmath69 and the symbol @xmath70 stands for @xmath8 if @xmath71 is true and @xmath9 if @xmath71 is false .
[ [ section-8 ] ] a model like [ model ] is often called a _
hierarchical _ model .
the @xmath72 and @xmath73 are called _ hyperparameters _ to distinguish them from the individual @xmath39 , @xmath45 and @xmath41 .
the model is fitted using bayesian inference with non - informative @xmath74 priors on the location parameters @xmath75 , @xmath76 and @xmath77 and inverse gamma @xmath78 priors on the scale parameters @xmath79 and @xmath73 . because intractable - looking integrals appear in the likelihood ( [ likelihood ] ) , the model is fitted using gibbs sampling . for the @xmath39 and @xmath41 ,
samples are drawn from the marginal distribution using a histogram approximation .
this is slow but easier to code than alternatives .
[ [ algorithm ] ] at each iteration of the algorithm , a value of @xmath80 is sampled using the theory of the normal distribution and then , for each character @xmath6 , the values of @xmath39 , @xmath45 and @xmath41 are sampled in that order
. then predictions for @xmath81 and @xmath82 are sampled using the definition of model [ model ] .
after all iterations are complete , a burn - in is discarded and the output is thinned to make the resulting samples as uncorrelated as possible .
this is useful for some purposes , such as drawing figure [ prob0 ] .
[ [ section-9 ] ] the output of the algorithm is a collection of samples of @xmath80 and predictions @xmath83 and @xmath84 where @xmath85 .
the algorithm is written in @xmath86 @xcite and uses the ` truncnorm ` package @xcite .
model [ model ] does not fit the training data very well since there are so many zeroes in column affc in table [ m ] .
it is known ( * ? ? ?
* afterword ) , ( * ? ? ?
* foreword ) that @xcite and @xcite were originally planned to be a single book , but it was later split into two volumes , each of which concentrates on a different subset of the characters .
[ [ preprocess ] ] this problem can be approached either by ignoring it , by modelling , or by pre - processing the data .
the model already has a lot of parameters and making it more complex is unlikely to be a good idea .
ignoring the problem and fitting the model to table [ m ] is not too bad , but it was decided to pre - process the data by replacing @xmath3 by @xmath87 where @xmath88 where @xmath89 and @xmath90 are the number of chapters in books @xmath91 and @xmath92 respectively .
this preserves the total number of chapters in each book , which may be of interest ( see section [ newcharacters ] . )
[ [ section-10 ] ] another possible approach would be to treat books @xmath91 and @xmath92 as one giant book when fitting the model .
the main disadvantage of this approach is that it decreases the amount of data available even further , although the resulting @xmath93 matrix would probably provide a better fit than @xmath87 to the chosen model .
the gibbs sampler of section [ algorithm ] was run for @xmath94 iterations with a burn - in of @xmath95 and thinned by taking every @xmath96 sample , resulting in posterior samples of size @xmath97 .
the algorithm was applied to the smoothed data @xmath87 of section [ datasmoothing ] .
it was run several times with random starting points to check that the results are stable .
only one run is recorded here .
[ [ section-11 ] ] table [ posteriors1 ] gives the posterior predictive distribution of pov chapters for each pov character in book @xmath10 .
table [ posteriors2 ] gives a similar distribution for book @xmath98 .
( the results for book @xmath98 are of less interest , as new predictions for book @xmath98 should be made following the appearance of book @xmath10 . )
graphs of the posterior distributions for book @xmath10 are given in figures [ posteriors_1 ] and [ posteriors_2 ] .
many of the distributions are bimodal and are not well - summarised by a single credible interval .
the distribution for tyrion has the highest variance , followed by jon snow ( see figure [ posteriors_1 ] . )
aeron , areo , jon connington , arianne@xmath99 , @xmath100quentyn , victarion , asha , barristan@xmath99 and @xmath100arys , melisandre@xmath99 the distributions are identical ; apparent differences are due to sampling variation . ]
one of the most compelling aspects of the _ song of ice and fire _ series is that major characters are frequently and unexpectedly killed off .
the probability of a character having zero pov chapters ( which is the first column of table [ posteriors1 ] divided by @xmath95 ) is therefore of interest .
these values are plotted in figure [ prob0 ] . treating the posterior samples as independent
, the error bars in figure [ prob0 ] indicate approximate @xmath101 confidence intervals of @xmath102 .
note that although a character who has been killed off will have zero pov chapters , the converse is not necessarily true .
the probabilities in figure [ prob0 ] are not based on events in the books , but solely reflect what the model can glean from table [ m ] . .
the characters are ordered on the @xmath1axis by the value of the posterior probability for book @xmath10 ( the black dots . )
the blue circles are the posterior probabilities of having zero pov chapters in book @xmath98 .
the error bars extend to @xmath102 and are intended to indicate when two posterior probabilities are roughly equal .
the figure is discussed in sections [ zeroprobabilities ] to [ isjonsnowdead ] . ]
[ [ section-12 ] ] the characters in figure [ prob0 ] are arranged on the @xmath1axis in order of their probability of having zero pov chapters in book @xmath10 .
eddard and catelyn were already killed off in @xcite and ( arguably ) @xcite respectively .
arys , who has the third highest posterior probability of having zero pov chapters , was killed off in @xcite , but it is misleading to be impressed by this . the reason why arys has a high posterior probability of having zero pov chapters is that he only ever appeared in one pov chapter , so @xmath103 is small and so , even if @xmath104 is large , there is a high probability that @xmath105 .
in fact , in the smoothed data , the rows for melisandre and arys are exactly the same .
the difference between the posterior distributions for these two characters is due to sampling variation ( and varies from one model run to another . )
[ [ section-13 ] ] the characters between tyrion and aeron in figure [ prob0 ] all have roughly the same posterior probability of having zero pov chapters .
they are mostly characters who have featured prominently in all the books since the beginning , together with the newer characters aeron , areo , arianne and jon connington whose posterior predictive distributions are identical and who have lower probabilities of non - appearance in book @xmath98 .
[ [ section-14 ] ] the next group of characters are those who have had relatively few pov chapters , including quentyn who , despite having been killed off in @xcite , is assigned a low posterior probability of @xmath106 of having zero pov chapters in book @xmath10 since he has the same posterior predictive distribution as asha , barristan and victarion .
[ [ section-15 ] ] finally , there are the characters cersei , brienne , jaime and samwell who have only recently become pov characters and have had a large number of pov chapters in the books in which they have appeared .
the model suggests that the probability of jon snow _ not _ being dead is at least @xmath107 since this is less than the posterior probability of his having at least one pov chapter in book @xmath10 .
given the events of @xcite , many readers would assess his probability of not being dead as being much lower than @xmath107 , but we must again point out that the model is unaware of the events in the books .
the model can only say that , based on the number of pov chapters observed so far , he has about as much chance of survival as the other major characters .
it is desirable to check that the model has been coded correctly .
a way to check this is to generate a data set according to the model and then see whether the chosen method of inference can recover the parameters which were used to generate the data set .
[ [ section-16 ] ] if the model had been fitted by frequentist methods , it would be possible to generate a large number of data sets , fit the model to each one , calculate confidence intervals for the hyperparameters , and check that the confidence intervals have the correct coverage .
since the model has been fitted by bayesian methods , it can only be used to produce credible intervals . however , as flat priors were used , the posterior distributions for the hyperparameters should be close to their likelihoods and so bayesian credible intervals should roughly coincide with frequentist confidence intervals when the posterior distributions of the hyperparameters are symmetric and unimodal , which they are .
[ [ inferencetest ] ] to test the method of inference , the model was fitted to @xmath108 data sets .
each data set was generated from hyperparameters which were a perturbation of @xmath109 .
these values were chosen because they were approximately the posterior medians for the hyperparameters obtained from one of the fits of model [ model ] to the smoothed data @xmath87 .
the location parameters @xmath110 and @xmath77 were perturbed by adding @xmath111 noise and the scale parameters @xmath79 and @xmath73 were perturbed by multiplying by @xmath112 where @xmath113 . for each of the data sets ,
@xmath114credible intervals were calculated by taking the middle @xmath115 of the posterior distributions for each of the @xmath10 hyperparameters , yielding @xmath116 credible intervals per @xmath114 .
the results , plotted in figure [ inference_test ] ( left panel ) indicate that the credible intervals have roughly the correct coverage .
[ inference_test ] model fits as described in section [ inferencetest ] and section [ inferencetest_pred ] . on the left , a comparison of the nominal and actual coverage of credible intervals for the hyperparameters . on the right , a comparison of the nominal and actual coverage of credible intervals for the predicted number of pov chapters.,title="fig : " ] [ [ section-17 ] ] note that there are choices of the hyperparameters which the chosen method of inference will not be able to recover .
for example , data generated with @xmath117 will be practically indistinguishable from data generated with @xmath118 , so there is no hope of inferring the hyperparameters in this case .
this is no great drawback as it should not affect the model s predictions , which are the topic of interest .
[ [ inferencetest_pred ] ] the procedure of section [ inferencetest ] was carried out for the one - step - ahead predictions for book @xmath10 .
the result , shown in figure [ inference_test ] ( right panel ) shows that the credible intervals have greater coverage than they should .
this is because the number of pov chapters can only take integer values and so the closed interval @xmath119 $ ] obtained by taking the @xmath120 and @xmath121 quantiles of the posterior distribution will in general cover more than @xmath115 of the posterior samples .
[ [ section-18 ] ] note once again that the purpose of these checks and experiments is to make sure that the model has been correctly coded .
we now discuss how to evaluate its predictions .
every predictive model should be applied to unseen test data to see how accurate its predictions really are .
it will not be possible to test model [ model ] before the publication of @xcite but an attempt at validation can be made by fitting the model to earlier books and seeing what it would have predicted for the next book .
[ [ validation_12 ] ] the model was tested by fitting it to books @xmath8 and @xmath122 in the series .
only @xmath14 pov characters appear in these books , so the data consist of the upper - left @xmath123 submatrix of table [ m ] .
figure [ validation ] shows the result of fitting the model to this matrix and comparing with the true values from the third column of table [ m ] .
the intervals displayed are central @xmath124 ( solid lines ) and @xmath21 ( dotted lines ) credible intervals .
the coverage is satisfactory but the intervals are much too wide to be of interest .
[ validation ] ( plotted as dots ) from table [ m ] compared with central @xmath124 ( solid lines ) and @xmath21 ( broken lines ) credible intervals obtained from fitting the model to @xmath125 as described in section [ validation_12 ] .
the characters have been sorted in increasing order of the posterior median.,title="fig : " ] [ [ section-19 ] ] fitting the model to the @xmath126 upper - left submatrix of table [ m ] consisting of pov chapters from the first three books gives more interesting output , but it is not clear how to evaluate the results because of the splitting of books @xmath91 and @xmath92 discussed above in section [ datasmoothing ] .
[ [ section-20 ] ] we can also compare the model s predictions with preview chapters from @xcite which are said to have been released featuring the points of view of arya , arianne , victarion and barristan .
given that there is at least one arya chapter , table [ posteriors1 ] indicates that there will probably be at least @xmath92 arya pov chapters and perhaps more .
[ [ section-21 ] ] the model can generate data containing a row of zeroes , but there are no zero rows in the data to which the model is fitted , because by definition this would correspond to a character who has never been a pov character in the books .
this is a source of bias in the model but it is not obvious how it can be avoided .
the effect of the bias can be tested by repeating the simulations of section [ inferencetest ] , but deleting zero rows before fitting the model . for @xmath127 ,
the coverage of a @xmath115 credible interval for a hyperparameter tends to be roughly @xmath128 .
there is little to support the choice of the poisson distribution in model [ model ] other than that it has the smallest possible number of parameters .
it is more common to use the negative binomial distribution for count data , but this would introduce extra complexity into the model , which is undesirable . with only @xmath0 values of @xmath39 , @xmath45 and @xmath41 available for finding the corresponding hyperparameters , it may not be possible to fit a ( truncated ) normal distribution in a meaningful way .
consideration of the posterior samples of the @xmath39 , @xmath45 and @xmath41 suggest that they more - or - less follow the pattern which is evident in the data and that the shrinkage of these parameters towards a common mean , which is one of the benefits of using a hierarchical model , can not really be attained with so little data .
for example , when the model is fitted to a data set containing a row in which the most recent entry is @xmath9 , the vast majority of posterior samples for the next entry in that row are always @xmath9 .
this is one reason for smoothing the data in section [ datasmoothing ] before fitting the model , in preference to fitting the model directly to table [ m ] .
if the number of pov characters was much larger , this might not be such a big problem .
given @xmath129 , the model treats @xmath31 and @xmath60 as independent for @xmath59 .
this is not a realistic assumption because if one character has more pov chapters , then the other characters will necessarily have fewer .
again , addressing this would seem to over - complicate the model .
the model ignores the introduction of new pov characters , although every book in the series has featured some new pov characters .
we can , however , use the output from the fitted model to make guesses about new characters .
the posterior distribution of the number of chapters in book @xmath10 told from the points of view of existing pov characters is unimodal with a mean of @xmath130 chapters , but typical books in the series so far have had about @xmath131 chapters .
so we could estimate that there will be about @xmath132 chapters in @xcite told from the point of view of new pov characters . in the previous books , according to table [ m ] , the number of chapters told from the point of view of new pov characters has been @xmath133 and @xmath132 , so @xmath132 does not seem like an unreasonable guess .
we could continue to make predictions in the hope of getting one right , but there is no merit in this .
we hope that it will be possible to review the model s performance following the publication of @xcite . | predictions are made for the number of chapters told from the point of view of each character in the next two novels in george r. r. martin s _ a song of ice and fire _ series by fitting a random effects model to a matrix of point - of - view chapters in the earlier novels using bayesian methods . *
spoiler warning : readers who have not read all five existing novels in the series should not read further , as major plot points will be spoiled , starting with table [ m ] . |
the sun is a variable star with an 11-year cyclic variation of its magnetic activity .
signatures of its variability can be seen in several activity proxies , the most famous being the number of spots on the surface of the sun .
other indices such as the 10.7-cm radio flux or the irradiance for instance also show this 11-year periodicity .
the 11-year magnetic cycle is theoretically understood to be driven by a magnetic dynamo process located at the bottom of the convective zone @xcite .
attempts have been made to provide prediction of the cycle properties based on dynamo models but these are not conclusive yet .
indeed , different conclusions were obtained for the new cycle 24 , for which these models had not predicted the long and deep minimum of cycle 23 . although spots on solar - type stars can not be directly observed , stellar magnetic cycles have been already reported .
indeed , the associated areas of concentrated magnetic field produce strong emission in the ca ii h and k spectral lines , which were proven to be a good proxy of surface magnetic fields @xcite .
surveys from the mount wilson and lowell observatories ( northern - hemisphere targets ) along with long - term monitoring campaigns with smarts 1.5-m telescope at ctio ( southern - hemisphere targets ) have revealed that many solar - type stars exhibit long - term cyclic variations in their ca ii h and k emission lines , analogous to the solar variations @xcite .
the complete sample includes cycle periods covering a range between 2.5 to 25 years .
recently , @xcite discovered an even shorter activity cycle of 1.6 years in the exoplanet hosting f8v star hd 17051 .
moreover , it has been observed that the periods of the activity cycles increase proportional to the stellar rotational periods along two distinct paths in main - sequence stars : the relatively young , active sequence , and the older , less active sequence @xcite .
relations between mean level of magnetic activity , rotation rate , and periods of the observed activity cycle generally support a dynamo interpretation .
however , spectroscopic measurements are only good proxies of surface magnetic fields .
the recent discovery , using asteroseismic observations , of variations with magnetic activity in the p - mode oscillation frequencies of the solar - like f5v star hd 49933 @xcite , with a frequency dependence comparable in shape to the one observed in the sun @xcite , opens a new era in the study of the physical phenomena involved in the dynamo processes .
indeed , through the precise measurements of oscillation parameters , the seismic observables provide unique information about the interior of the stars and for the determination of key parameters for the study of stellar activity , such as the depth of the convection zone , the characteristic evolution time of the granulation , the differential rotation , or the sound - speed as a function of the star s radius ( see the preliminary work on hd 49933 by * ? ? ? * ) .
all these new observables will impose new constraints to the theory ( e.g. * ? ? ? * ; * ? ? ?
long , high - quality photometric observations such as the ones collected by the space - based convection , rotation , and planetary transits ( corot , * ? ? ?
* ) and _ kepler _
@xcite missions will contribute to probe stellar cycles in a wide variety of solar - type stars .
the success of helioseismology is due to its capability to accurately measure the p - mode parameters of the solar eigenmode spectrum , which allow us to infer unique information about the internal structure and dynamics of the sun from its surface all the way down to the core .
it has contributed greatly to a clearer understanding of the sun and provided insights into the complex solar magnetism , by means for instance of the variability of the characteristics of the p - mode spectrum .
evidence of p - mode frequency changes with solar activity , first revealed by @xcite , were established by @xcite with the analysis of helioseismic observations spanning the complete solar cycle 21 ( 19771988 ) . as longer , higher quality , and continuous helioseismic observations became available , the solar p - mode frequencies proved to be very sensitive to the solar surface activity ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) with high levels of correlation with solar surface activity proxies . the low - degree , p - mode frequencies change by about @xmath0hz between the minimum and the maximum of the solar cycle with correlation levels with surface activity proxies higher than 0.9
moreover , @xcite showed that the p - mode frequencies are shifted in presence of surface magnetic activity , varying with close temporal and spatial distributions .
however , clear differences were observed between the frequency shifts and the surface activity of the sun during the unusually extended and deep minimum of cycle 23 @xcite .
although the form and the dependence of the shifts indicate a near - surface phenomenon explained to arise from changes in the outer layers of the sun , the origin of the frequency shifts is far from being properly understood . moreover , a quasi - biennial signal in the solar frequencies was recently observed by @xcite , indicating the possible action of a second dynamo seated near the bottom of the layer extending 5% below the solar surface .
the other p - mode oscillation parameters such as the amplitude , the lifetime , and the asymmetry , were also observed to vary with solar activity @xcite .
for example , the mode amplitudes decrease with increasing solar activity by about 40% from minimum to maximum of the 11-year cycle .
the temporal variations between frequency and amplitude are therefore anticorrelated .
the solar - like f5v star hd 49933 observed by the corot satellite @xcite is the first star after the sun for which variability of the p - mode parameters with magnetic activity was observed @xcite .
indeed , the p - mode frequencies and amplitudes vary with time with a clear anti - correlation between both parameters as observed in the sun ( e.g. * ? ? ?
these temporal variations suggest a cycle period of at least 120 days .
preliminary spectroscopic measurements with of the ca ii h and k emission lines had shown that hd 49933 is an active star , with a mount wilson s - index of 0.3 .
follow - up long - term monitoring confirms the existence of a short activity cycle ( t. metcalfe , private communication , 2011 ) .
incoming observations ( both asteroseismic and spectroscopic ) will help us to determine more accurately the period and the properties of this magnetic activity cycle .
@xcite showed that the frequency shifts measured in hd 49933 present a frequency dependence with a clear increase with frequency , reaching a maximal shift of about @xmath1hz around @xmath2hz , which shows a similar pattern as in the sun .
that indicates then the presence of similar physical phenomena driving the frequency shifts of the oscillation modes as the ones taking place in the sun , which are understood to reflect structural changes in and just below the photosphere with stellar activity if we suppose similar mechanisms as in the sun ( e.g. , * ? ? ?
* ) . however , the frequency shift measured in hd 49933 is at least five times larger than in the sun , which reaches about @xmath3hz at @xmath4hz between the maximum and the minimum of the 11-year solar cycle ( e.g. * ? ? ? * ) .
this observation supports the scaling proposed by @xcite , who predicted that stars hotter and more evolved than the sun ( like hd 49933 ) should have larger frequency shifts than in the sun .
preliminary structure models of hd 49933 were computed by @xcite in order to study the effects of sound - speed perturbations in the near surface layers on the p - mode frequencies .
this will provide for instance insights on the properties of the convective zone of hd 49933 and on the depth at which the magnetic field perturbs the modes .
three other solar - like stars observed by corot have been studied , for which spectropolarimetric measurements from the narval instrument @xcite located at the pic du midi observatory are available : hd 181420 @xcite , hd 49385 @xcite , and hd 52265 @xcite .
although hd 181420 and hd 49385 seem to be in a quiescence state , hd 52265 shows small temporal variations of the p - mode parameters , suggesting a modest increase of magnetic activity , also indicated by the spectroscopic observations @xcite .
asteroseismology provides not only invaluable information to infer the structure and the dynamics of the interior of the stars but also key parameters for the study of stellar magnetic activity cycles in a wide variety of solar - type stars imposing new observational constraints to the theory .
studies of stellar activity cycles will bring important inferences for the modeling of dynamo processes and will put the sun and its 11-year activity cycle into context .
the solar - like f star , hd 49933 , is the first star after the sun for which variations of the oscillation parameters with magnetic activity have been observed .
although these changes follow analogous patterns as in the sun , they suggest a short cycle period , which is confirmed by spectroscopic observations . they also support that stars hotter and more evolved than the sun should have larger frequency shifts .
this is important for corot and _ kepler _ when searching for similar frequency shifts as it suggests that the shifts should be easier to detect in the f stars .
also , as short activity cycles seem to be more common than expected , the corot and _ kepler _ ( and future ) missions should potentially be able to observe a large number of full cycles .
such studies will also provide inputs to explore the impacts of magnetic activity on possible planets hosted by these stars .
baglin , a. , michel , e. , auvergne , m. , & the corot team 2006 , in proceedings of soho 18/gong 2006/helas i , beyond the spherical sun , esa sp-624 , ed .
k. fletcher , & m. thompson ( esa special publication , noordwijk ) , 34 ceillier , t. , ballot , j. , garca , r. a. , et al .
2011 , in proceedings of the 2@xmath5 corot symposium , transiting planets , vibrating stars , and their connection , ed .
a. baglin , m. deleuil , e. michel , & c. moutou , to be published | the success of helioseismology is due to its capability to accurately measure the p - mode parameters of the solar eigenmode spectrum , which allow us to infer unique information about the internal structure and dynamics of the sun from its surface all the way down to the core .
it has contributed greatly to a clearer understanding of the sun and provided insights into the complex solar magnetism , by means for instance of the variability of the characteristics of the p - mode spectrum .
indeed , variations in the mean strength of the solar magnetic field lead to significant shifts in the frequencies of even the lowest - degree p modes with high levels of correlation with solar surface activity proxies .
these frequency shifts are explained to arise from structural changes in the outer layers of the sun during the 11-year activity cycle , which is understood to be driven by a dynamo process .
however , clear differences between p - mode frequencies and solar surface activity during the unusually extended minimum of cycle 23 were observed .
the origin of the p - mode variability is thus far from being properly understood and a better comprehension of its relationship with solar and stellar activity cycles will help us in our understanding of the dynamo processes .
spectroscopic measurements of ca h and k emission lines revealed magnetic activity variations in a large sample of solar - type stars with timescales ranging from 2.5 and 25 years .
this broad range of cycle periods is thought to reflect differences in the rotational properties and the depths of the surface convection zones with various masses and ages
. however , spectroscopic measurements are only good proxies of surface magnetic fields .
the recent discovery of variations with magnetic activity in the p - mode oscillation frequencies of the solar - like star hd 49933 observed by corot , with a frequency dependence comparable in shape to the one observed in the sun , opens a new era in the study of the physical phenomena involved in the dynamo processes .
current and future asteroseismic observations will contribute to probe stellar cycles in a wide variety of solar - type stars .
solar - type stars , activity , oscillations |
consider the problem of finding a closed hypersurface of prescribed curvature @xmath0 in a globally hyperbolic lorentzian manifold @xmath1 having a compact cauchy hypersurface @xmath2 . to be more precise ,
let @xmath3 be a connected open subset of @xmath4 a smooth , symmetric function defined in an open cone @xmath5 $ ] , then we look for a space - like hypersurface @xmath6 such that @xmath7 where @xmath8 means that @xmath0 is evaluated at the vector @xmath9 the components of which are the principal curvatures of @xmath10 .
the prescribed function @xmath11 should satisfy natural structural conditions , e. g. if @xmath12 is the positive cone and the hypersurface @xmath10 is supposed to be convex , then @xmath11 should be positive , but no further , merely technical , conditions should be imposed . in the case
@xmath13 , the mean curvature , has been treated , and in we solved the problem for curvature functions @xmath0 of class @xmath14 that includes the gaussian curvature , see for the definition , but excludes the symmetric polynomials @xmath15 for @xmath16 . among these
, @xmath17 , that corresponds to the scalar curvature operator , is of special interest . however , a solution of equation with @xmath18 is in general not a hypersurface of prescribed scalar curvature unless the ambient space has constant curvature since the scalar curvature of a hypersurface also depends on @xmath19 .
thus , we have to allow that the right - hand side @xmath11 also depends on time - like vectors and look for hypersurfaces @xmath10 satisfying @xmath20 where @xmath21 is the past - directed normal of @xmath10 in the point @xmath22 . to give a precise statement of the existence result we need a few definitions and assumptions .
first , we assume that @xmath3 is a precompact , connected , open subset of @xmath1 , that is bounded by two _ achronal _ , connected , space - like hypersurfaces @xmath23 and @xmath24 of class @xmath25 , where @xmath23 is supposed to lie in the past of @xmath24 .
let @xmath18 be the scalar curvature operator defined on the open cone @xmath26 $ ] , and @xmath27 be of class @xmath28 in its arguments such that @xmath29 \nnorm{f_\bet(x,\n)}&\le c_2 ( 1+\nnorm\n^2),\lae{0.4}\\ \intertext{and } \nnorm{f_{\n^\bet}(x,\n)}&\le c_3 ( 1+\nnorm\n),\lae{0.5}\end{aligned}\ ] ] for all @xmath30 and all past directed time - like vectors @xmath31 , where @xmath32 is a riemannian reference metric that will be detailed in .
we suppose that the boundary components @xmath33 act as barriers for @xmath34 .
@xmath24 is an _
upper barrier _ for @xmath34 , if @xmath24 is _ admissible _ ,
i.e. its principal curvatures @xmath35 with respect to the past directed normal belong to @xmath36 , and if @xmath37 @xmath23 is a lower barrier for @xmath34 , if at the points @xmath38 , where @xmath23 is admissible , there holds @xmath39 @xmath40 may be empty .
this definition of upper and lower barriers for a pair @xmath34 also makes sense for other curvature functions @xmath0 defined in an open convex cone @xmath41 , with a corresponding meaning of the notion _
admissable_. now , we can state the main theorem .
let @xmath23 be a lower and @xmath24 an upper barrier for @xmath34 , where @xmath18 .
then , the problem @xmath42 has an admissible solution @xmath43 of class @xmath25 that can be written as a graph over @xmath44 provided there exists a strictly convex function @xmath45 .
as we have shown in the existence of a strictly convex function @xmath46 is guaranteed by the assumption that the level hypersurfaces @xmath47 are strictly convex in @xmath48 , where @xmath49 is a gaussian coordinate system associated with @xmath2 .
looking at robertson - walker space - times it seems that the assumption of the existence of a strictly convex function in the neighbourhood of a given compact set is not too restrictive : in minkowski space e.g. @xmath50 is a globally defined strictly convex function .
the only obstruction we are aware of is the existence of a compact maximal slice . in the neighbourhood of such a slice
a strictly convex function can not exist .
the existence result of our main theorem would also be valid in riemannian manifolds if one could prove @xmath51- estimates .
for the @xmath52- estimates the nature of the ambient space is irrelevant though the proofs are slightly different . for prescribed curvature problems it seems more natural to assume that the right - hand side @xmath11 depends on @xmath53 , and we shall prove in a subsequent paper existence results for curvature functions @xmath54 , where the ambient space can be riemannian or lorentzian , cf . .
the paper is organized as follows : in 1 we take a closer look at curvature functions and define the concept of _ elliptic regularization _ for these functions , and analyze some of its properties . in 2
we introduce the notations and common definitions we rely on , and state the equations of gau , codazzi , and weingarten for space - like hypersurfaces . in 3
we look at the curvature flow associated with our problem , and the corresponding evolution equations for the basic geometrical quantities of the flow hypersurfaces . in 4
we prove lower order estimates for the evolution problem , while a priori estimates in the @xmath52 are derived in 5 .
in 6 , we demonstrate that the evolutionary solution converges to a stationary approximation of our problem , i.e. to a solution for a curvature problem , where @xmath0 is replaced by its elliptic regularization @xmath55 .
the uniform @xmath51- estimates for the stationary approximations are derived in sections [ s:7 ] and [ s:8 ] , the @xmath52- estimates are given in , while the final existence result is contained in .
let @xmath5 $ ] be an open cone containing the positive cone @xmath56 , and @xmath57 a positive symmetric function satisfying the condition @xmath58 then , @xmath0 can also be viewed as a function defined on the space of symmetric matrices @xmath59 , the eigenvalues of which belong to @xmath41 , namely , let @xmath60 with eigenvalues @xmath61 , then define @xmath0 on @xmath62 by @xmath63 if we define @xmath64 then , @xmath65,\ ] ] in an appropriate coordinate system , @xmath66 and @xmath67 for any @xmath68 , where @xmath69 is the space of all symmetric matrices .
the second term on the right - hand side of is non - positive if @xmath0 is concave , and non - negative if @xmath0 is convex , and has to be interpreted as a limit if @xmath70 .
the preceding considerations are also applicable if the @xmath71 are the principal curvatures of a space - like hypersurface @xmath10 with metric @xmath72 .
@xmath0 can then be looked at as being defined on the space of all symmetric tensors @xmath73 the eigenvalues of which belong to @xmath41 .
such tensors will be called ; when the second fundamental form of @xmath10 is admissible , then , we also call @xmath10 admissible . for an admissible tensor @xmath73
@xmath74 is a contravariant tensor of second order
. sometimes it will be convenient to circumvent the dependence on the metric by considering @xmath0 to depend on the mixed tensor @xmath75 then , @xmath76 is also a mixed tensor with contravariant index @xmath77 and covariant index @xmath78 .
such functions @xmath0 are called curvature functions .
important examples are the symmetric polynomials of order @xmath79 , @xmath15 , @xmath80 , @xmath81 they are defined on an open cone @xmath82 that can be characterized as the connected component of @xmath83 that contains @xmath56 .
since we have in mind that the @xmath71 are the principal curvatures of a hypersurface , we use the standard symbols @xmath84 and @xmath85 for @xmath86 the scalar curvature function @xmath18 can then be expressed as @xmath87 and we deduce that for @xmath88 @xmath89 f_i&=h-\ka_i,\lae{1.16}\\ \intertext{and hence , } h f_i&\ge f,\lae{1.17}\end{aligned}\ ] ] for is equivalent to @xmath90 which is obviously valid . in important ingredient in our existence proof
will be the method of _ elliptic regularization _ for each @xmath91
, consider the linear isomorphism @xmath92 in @xmath93 $ ] given by @xmath94 let @xmath95 be a curvature function such that @xmath96 then , @xmath97 is an open cone and @xmath98 a curvature function satisfying @xmath99 assume furthermore , that @xmath100 then , @xmath101 we only prove the assertions and since the other assertions are obvious .
let @xmath102 be fixed .
then , @xmath103 because @xmath0 is monotone , and we deduce @xmath104 in view of and the monotonicity of @xmath0 , cf . .
to prove , we observe that @xmath105 \(i ) let @xmath0 be as in and assume moreover , that @xmath0 is homogeneous of degree @xmath106 , and concave , then , @xmath107 and we conclude that condition is satisfied .
\(ii ) let @xmath0 be as in , but suppose that @xmath0 is homogeneous of degree @xmath108 and @xmath109 concave , then , the relation is also valid .
the inequality follows easily from the concavity and homogeneity @xmath110 & = \frac1n f(1,\dotsc , 1)h , \end{aligned}\ ] ] since @xmath111 , while the other assertions are obvious . for better reference
, we use a tensor setting in the next lemma , i.e. the @xmath102 are the eigenvalues of an admissible tensor @xmath73 with respect to a riemannian metric @xmath72 . in this setting the elliptic regularization of @xmath0 is given by @xmath112 let @xmath113 be the elliptic regularization of a curvature function @xmath0 of class @xmath52 , then , @xmath114 and @xmath115 & \hp{=\ ; } + \e f^{rs , kl } g_{rs } g^{ij } + \e^2 f^{rs , ab } g_{rs } g_{ab } g^{ij } g^{kl}. \end{aligned}\ ] ] if @xmath0 is concave , then , @xmath113 is also concave .
the relations and are straight - forward calculations . to prove the concavity of @xmath113 , let @xmath116 be a symmetric tensor , then , @xmath117 & \hp{=\ ; } + \e^2 f^{rs , ab } g_{rs } g_{ab } ( g^{ij}\h_{ij})^2\le 0 .
\end{aligned}\ ] ]
the main objective of this section is to state the equations of gau , codazzi , and weingarten for space - like hypersurfaces @xmath10 in a lorentzian space @xmath1 .
geometric quantities in @xmath1 will be denoted by @xmath118 , etc . , and those in @xmath10 by @xmath119 , etc .
greek indices range from @xmath120 to @xmath121 and latin from @xmath106 to @xmath121 ; the summation convention is always used .
generic coordinate systems in @xmath1 resp .
@xmath10 will be denoted by @xmath49 resp .
covariant differentiation will simply be indicated by indices , only in case of possible ambiguity they will be preceded by a semicolon , i.e. for a function @xmath123 in @xmath1 , @xmath124 will be the gradient and @xmath125 the hessian , but e.g. , the covariant derivative of the curvature tensor will be abbreviated by @xmath126 .
we also point out that @xmath127 with obvious generalizations to other quantities .
let @xmath10 be a hypersurface , i.e. the induced metric is riemannian , with a differentiable normal @xmath128 that is time - like . in local coordinates , @xmath49 and @xmath122 ,
the geometric quantities of the space - like hypersurface @xmath10 are connected through the following equations @xmath129 the so - called . here , and also in the sequel ,
a covariant derivative is always a tensor , i.e. @xmath130 the comma indicates ordinary partial derivatives . in this implicit definition
the @xmath73 is taken with respect to @xmath128 .
the second equation is the @xmath131 where we remember that @xmath132 is a full tensor .
finally , we have the @xmath133 and the @xmath134 now , let us assume that @xmath1 is a globally hyperbolic lorentzian manifold with a cauchy surface .
@xmath1 is then a topological product @xmath135\times \mc s_0 $ ] , where @xmath44 is a compact riemannian manifold , and there exists a gaussian coordinate system @xmath49 , such that @xmath136 represents the time , the @xmath137 are local coordinates for @xmath44 , where we may assume that @xmath44 is equal to the level hypersurface @xmath138we do nt distinguish between @xmath44 and @xmath139 , and such that the lorentzian metric takes the form @xmath140 where @xmath141 is a riemannian metric , @xmath142 a function on @xmath1 , and @xmath22 an abbreviation for the space - like components @xmath143 , see , , , and .
we also assume that the coordinate system is , i.e. the time coordinate @xmath136 increases on future directed curves .
hence , the time - like vector@xmath144 is future directed as is its version @xmath145 .
let @xmath146 be a space - like hypersurface @xmath147 then the induced metric has the form @xmath148 where @xmath141 is evaluated at @xmath149 , and its inverse @xmath150 can be expressed as @xmath151 where @xmath152 and @xmath153 hence , @xmath154 is space - like if and only if @xmath155 .
we also note that @xmath156 the covariant form of a normal vector of a graph looks like @xmath157 and the contravariant version is @xmath158 thus , we have let @xmath10 be space - like graph in a future oriented coordinate system .
then , the contravariant future directed normal vector has the form @xmath159 and the past directed @xmath160 in the gau formula we are free to choose the future or past directed normal , but we stipulate that we always use the past directed normal for reasons that we have explained in .
look at the component @xmath161 in and obtain in view of @xmath162 here , the covariant derivatives are taken with respect to the induced metric of @xmath10 , and @xmath163 where @xmath164 is the second fundamental form of the hypersurfaces @xmath165 .
an easy calculation shows @xmath166 where the dot indicates differentiation with respect to @xmath136 .
next , let us analyze under which condition a space - like hypersurface @xmath10 can be written as a graph over the cauchy hypersurface @xmath44 .
we first need let @xmath10 be a closed , space - like hypersurface in @xmath1 .
then , @xmath10 is said to be , if no two points in @xmath10 can be connected by a future directed time - like curve . in it
is proved , see also , let @xmath1 be connected and globally hyperbolic , @xmath167 a compact cauchy hypersurface , and @xmath168 a compact , connected space - like hypersurface of class @xmath169 .
then , @xmath170 with @xmath171 iff @xmath10 is achronal .
the @xmath33 are barriers for the pair @xmath34 .
let us point out that without loss of generality we may assume @xmath172 for let @xmath173 be a function with support in a small neighbourhood of @xmath174the dot should indicate that the union is disjoint such that @xmath175 and define for @xmath176 @xmath177 then , if we assume @xmath11 to be strictly positive with a positive lower bound , we have for small @xmath178 @xmath179 and the @xmath33 are barriers for @xmath180 satisfying the strict inequalities ; since we shall derive @xmath25 estimates independent of @xmath178 , we shall have proved the existence of a solution for @xmath11 if we can prove it for @xmath181 .
let @xmath33 be barriers for @xmath34 satisfying the strict inequalities and , where @xmath0 is supposed to be monotone and concave .
then , they are also barriers for the elliptic regularizations @xmath55 for small @xmath182 . in view of , we know that @xmath183 and @xmath84 is positive in @xmath184 . hence , @xmath24 is certainly an upper barrier for @xmath185 because of the monotonicity of @xmath0 .
let @xmath186 resp .
@xmath40 be the points in @xmath23 where the principal curvatures belong to @xmath184 resp . @xmath41 and assume that @xmath187 .
suppose @xmath23 were not a lower barrier for @xmath185 for small @xmath182 , then , there exist a sequence @xmath188 and a corresponding convergent sequence @xmath189 , @xmath190 , such that @xmath191 and hence , @xmath192 contradicting .
the condition is reasonable as is evident from the einstein equation @xmath193 where the energy - momentum tensor @xmath194 is supposed to be positive semi - definite for time - like vectors ( _ weak energy condition _
, cf . ) , and the relation @xmath195+\bar r+2 \bar r_{\al\bet}\n^\al\n^\bet\ ] ] for the scalar curvature of a space - like hypersurface ; but it would be convenient for the approximations we have in mind , if the estimate in would be valid for all time - like vectors . in fact , we may assume this without loss of generality : let @xmath196 be a smooth real function such that @xmath197 then , we can replace @xmath11 by @xmath198 and the new function satisfies our requirements for all time - like vectors .
we therefore assume in the following that the relation holds for all time - like vectors @xmath199 and all @xmath200 . sometimes , we need a riemannian reference metric , e.g. if we want to estimate tensors .
since the lorentzian metric can be expressed as @xmath201 we define a riemannian reference metric @xmath202 by @xmath203 and we abbreviate the corresponding norm of a vectorfield @xmath204 by @xmath205 with similar notations for higher order tensors . for a space - like hypersurface @xmath206 the induced metrics with respect to @xmath207 resp .
@xmath208 are related as follows @xmath209\\[\cma ] & = g_{ij}+2 e^{2\psi } u_i u_j . \end{aligned}\ ] ] thus , if @xmath210 is a unit vector for @xmath72 , then @xmath211 and we conclude for future reference let @xmath206 be a space - like hypersurface in @xmath1 , @xmath212 , and @xmath213 a unit vector , then @xmath214 where @xmath215 .
solving the problem involves two steps : first , proving a priori estimates , and secondly , applying a method to show the existence of a solution . in a general
lorentzian manifold the evolution method is the method of choice , but unfortunately , one can not prove the necessary a priori estimates during the evolution when @xmath0 is the scalar curvature operator .
both the @xmath51 and @xmath52- estimates fail for general @xmath27 .
therefore , we use the and consider the existence problem for the operators @xmath216 i.e. we solve @xmath217 then , we prove uniform @xmath28- estimates for the approximating solutions @xmath218 , and finally , let @xmath182 tend to zero .
the @xmath55or some positive power of it belong to a class of curvature functions @xmath0 that satisfy the following condition @xmath219 : @xmath220 , where @xmath221 $ ] is an open cone containing @xmath56 , @xmath0 is symmetric , monotone , i.e. @xmath222 , homogeneous of degree 1 , concave , vanishes on @xmath223 , and there exists @xmath224 such that @xmath225 furthermore , the set @xmath226 is compact . if the original curvature function @xmath227 is concave , homogeneous of degree 1 , and vanishes on @xmath228 , then , the @xmath55 are of class @xmath219 in the cone @xmath184 , and satisfy with @xmath229 .
the set @xmath230 is compact for fixed @xmath182 . if the parameters @xmath231 and @xmath178 are independent of @xmath182 , then the @xmath232 are contained in a compact subset of @xmath41 uniformly in @xmath182 , for small @xmath182 , @xmath233 . in view of the results in we only have to prove the compactness of @xmath232 .
we shall also only consider the case when the estimates hold uniformly in @xmath182 . due to the concavity and homogeneity of @xmath55 we conclude from that @xmath234 for @xmath235 we , therefore , infer @xmath236 \intertext{and thus , } \lim_{\e\ra 0}\e h&=0,\lae{3.7b}\end{aligned}\ ] ] uniformly in @xmath237 .
suppose @xmath237 would not stay in a compact subset of @xmath41 for small @xmath238 .
then , there would exist a sequence @xmath239 and a corresponding sequence @xmath240 converging to a point @xmath241 , which is impossible in view of , , and the continuity of @xmath0 in @xmath242 . to prove the existence of hypersurfaces of prescribed curvature @xmath0 for @xmath243 we look at the evolution problem @xmath244 x(0)&=x_0 , \end{aligned}\ ] ] where @xmath128 is the past - directed normal of the flow hypersurfaces @xmath245 , @xmath0 the curvature evaluated at @xmath245 , @xmath246 an embedding and @xmath247 an embedding of an initial hypersurface @xmath248 , which we choose to be the upper barrier @xmath24 . since @xmath0 is an elliptic operator , short - time existence , and hence , existence in a maximal time interval @xmath249 is guaranteed .
if we are able to prove uniform a priori estimates in @xmath28 , long - time existence and convergence to a stationary solution will follow immediately .
but before we prove the a priori estimates , we want to show how the metric , the second fundamental form , and the normal vector of the hypersurfaces @xmath245 evolve .
all time derivatives are derivatives .
the proofs are identical to those of the corresponding results in a riemannian setting , cf . and , and will be omitted .
the metric @xmath250 of @xmath245 satisfies the evolution equation @xmath251 the normal vector evolves according to @xmath252 the second fundamental form evolves according to @xmath253 and @xmath254 the term @xmath255 evolves according to the equation @xmath256- f^{ij}h_{ik}h_j^k ( f- f ) - f_\al\n^\al ( f- f)\\-f_{\n^\al}x^\al_i(f - f)_jg^{ij } - f^{ij}\riema \al\bet\ga\de\n^\al x_i^\bet \n^\ga x_j^\de ( f- f),\end{gathered}\ ] ] from we deduce with the help of the ricci identities a parabolic equation for the second fundamental form the mixed tensor @xmath257 satisfies the parabolic equation @xmath258 the proof is identical to that of the corresponding result in the riemannian case , cf .
; the only difference is that @xmath11 now also depends on @xmath128 . in view of the maximum principle
, we immediately deduce from that the term @xmath255 has a sign during the evolution if it has one at the beginning , i.e. , if the starting hypersurface @xmath248 is the upper barrier @xmath24 , then @xmath255 is non - negative @xmath259
since the two boundary components @xmath260 of @xmath261 are space - like , achronal hypersurfaces , they can be written as graphs over the cauchy hypersurface @xmath2 , @xmath262 , @xmath263 , and we have @xmath264 for @xmath23 should lie in the past of @xmath24 , and the enclosed domain is supposed to be connected .
let us look at the evolution equation with initial hypersurface @xmath248 equal to @xmath24 defined on a maximal time interval @xmath265 .
since the initial hypersurface is a graph over @xmath2 , we can write @xmath266 where @xmath123 is defined in the cylinder @xmath267 .
we then deduce from , looking at the component @xmath161 , that @xmath123 satisfies a parabolic equation of the form @xmath268 where we use the notations in , and where we emphasize that the time derivative is a total derivative , i.e. @xmath269 since the past directed normal can be expressed as @xmath270 we conclude from , , and @xmath271 thus , @xmath272 is non - positive in view of .
next , let us state our first a priori estimate suppose that the boundary components act as barriers for @xmath34 , then the flow hypersurfaces stay in @xmath273 during the evolution .
the proof is identical to that of the corresponding result in . for the @xmath51- estimate the term @xmath215 is of great importance .
it satisfies the following evolution equation consider the flow in the distinguished coordinate system associated with @xmath2 .
then , @xmath274 satisfies the evolution equation @xmath275 where @xmath204 is the covariant vector field @xmath276 .
the proof uses the relation @xmath277 and is identical to that of having in mind that presently @xmath11 also depends on @xmath128 .
let @xmath278 be the flow hypersurfaces , then , we have @xmath279 where all covariant derivatives a taken with respect to the induced metric of the flow hypersurfaces , and the time derivative @xmath280 is the total time derivative , i.e. it is given by .
we use the relation together with . as an immediate consequence
we obtain the composite function @xmath281 where @xmath282 are constants , satisfies the equation @xmath283 e^{\lam u } \,\f + f^{ij } u_i u_j \,\cha000\,\m
\lam\,e^{\lam u}\f\\[\cma ] & + 2\msp f^{ij } u_i \cha 0j0\,\m \lam\msp[2 ] e^{\lam u}\,\f+ f^{ij}\cha ij0\,\m\lam\msp[2 ] e^{\lam u}\,\f\\[\cma ] & -[1+\m\msp[2 ] e^{\lam u } ] f^{ij } u_i u_j\,\m \lam^2\msp[2 ] e^{\lam u}\,\f . \end{aligned}\ ] ] before we can prove the @xmath51- estimates we need two more lemmata .
there is a constant @xmath284 such that for any positive function @xmath285 on @xmath2 and any hypersurface @xmath245 of the flow we have @xmath286 g^{ij}&\le c\msp \tilde v^2\msp \s^{ij},\lae{4.12}\\[\cma ] f^{ij}&\le f^{kl } g_{kl}\msp g^{ij},\lae{4.13}\end{aligned}\ ] ] @xmath287 @xmath288 and @xmath289 \(i ) the first three inequalities are obvious .
\(ii ) follows from the generalized schwarz inequality combined with and .
\(iii ) is a direct consequence of and .
\(iv ) the proof of is a bit more complicated and uses the symmetry properties of the riemann curvature tensor .
let @xmath290 we shall show that the symmetrization of @xmath291 satisfies @xmath292 with a uniform constant @xmath284 , which in turn yields .
let @xmath293 be arbitrary , @xmath49 be the special gaussian coordinate of @xmath1 , and @xmath122 local coordinates around @xmath294 such that @xmath295 we also note that all indices are raised with respect to @xmath296 with the exception of the contravariant vector @xmath297 we point out that @xmath298 \tilde v^2&=1+e^{2\psi } \norm{du}^2,\lae{4.22}\\[\cma ] ( \n^\al)&=- \tilde v(1 , \check u^i ) e^{-\psi},\\\intertext{and } \h_\e x^\e_l g^{kl}&=-e^\psi u^k.\lae{4.24}\end{aligned}\ ] ] first , let us observe that in view of and the symmetry properties of the riemann curvature tensor we have @xmath299 next , we shall expand the right - hand side of explicitly .
@xmath300 & \hp{=}+ \riema l0k0 \tilde vu^k \check u^l u_i u_j + \riema l00j \tilde v \check u^l u_i \norm{du}^2\\[\cma ] & \hp{= } + \riema l0kj \tilde v u^k\check u^l u_i + \riema li0j \tilde v \check u^l\norm{du}^2\\[\cma ] & \hp{= } + \riema lik0 \tilde v u^k \check u^lu_j + \riema likj \tilde v u^k \check u^l \end{aligned}\ ] ] to prove the estimate , we may assume that @xmath301 .
let @xmath302 , @xmath303 be an orthonormal base of @xmath304 such that @xmath305 then , for @xmath306 , the @xmath307 are also orthonormal with respect to the metric @xmath308 , and it is also valid that @xmath309 where @xmath310 . for @xmath311
we deduce from @xmath312 & \hp{= } + \riema li0j\tilde v \check u^l\norm{du}^2 e^i_r e^j_s+ \riema likj \tilde v u^k \check u^l e^i_r e^j_s \end{aligned}\ ] ] and hence , @xmath313 it remains to estimate @xmath314 for @xmath315 , because of .
we deduce from @xmath316 where we used the symmetry properties of the riemann curvature tensor .
hence , we conclude @xmath317 and the relation is proved .
let @xmath318 be a graph over @xmath2 , @xmath206 , and @xmath319 a function given in @xmath2 , @xmath320 .
let @xmath321 be defined through @xmath322 where @xmath323 and @xmath324 .
then , there exists @xmath284 such that @xmath325 & \hp{\le } + \frac1{1 - 2\e } f^{ij } u_i u_j \m^2 \lam^2 e^{2\lam u } \tilde v \f .
\end{aligned}\ ] ] since @xmath326 , we have @xmath327 & = \h_{\al\bet } \n^\al x^\bet_i -e^\psi h^k_i u_k . \end{aligned}\ ] ] thus , we derive @xmath328 & \le c\msp f^{ij } g_{ij } \tilde v^3 \abs\lam \m e^{\lam u } \f + 2 e^\psi \abs{f^{ij } h^k_i u_k u_j}\abs\lam \m e^{\lam u } \f .
\end{aligned}\ ] ] the last term of the preceding inequality can be estimated by @xmath329 and we obtain the desired estimate in view of . applying to the evolution equation for @xmath330
we conclude there exists a constant @xmath284 such that for any function @xmath182 , @xmath331 , defined on @xmath2 the term @xmath330 satisfies an evolution inequality of the form @xmath332 & \hp{\le } + \frac{c}\e f^{ij } g_{ij } \tilde v^3 + c\msp \nnorm{f_\bet } \tilde v^2 + f_{\n^\bet } x^\bet_l h^{kl } u_k e^\psi . \end{aligned}\ ] ] we are now ready to prove the uniform boundedness of @xmath330 .
assume that there are positive constants @xmath333 , @xmath334 , such that for any @xmath335 and any past directed time - like vector @xmath128 there holds @xmath336
\nnorm{f_\bet(x,\n)}&\le c_2 ( 1+\nnorm{\n}),\lae{4.40}\\ \intertext{and } \nnorm{f_{\n^\bet}(x,\n)}&\le c_3.\lae{4.41}\end{aligned}\ ] ] then , the term @xmath330 remains uniformly bounded during the evolution @xmath337 where @xmath338 is the constant in . here , and in the following
, the reference that a constant depends on @xmath3 also means that it depends on the barriers and geometric quantities of the ambient space restricted to @xmath3 .
we proceed similar as in and show that the function @xmath339 @xmath321 as in , is uniformly bounded , if we choose @xmath340 appropriately , and assume furthermore , without loss of generality , that @xmath341 , for otherwise replace @xmath123 by @xmath342 , @xmath343 large , in the definition of @xmath321 . with the help of the lemmata [ l:4.4 ] , [ l:4.6 ] , and [ l:4.7 ] we derive from the relation @xmath344 \f+ [ \dot\f - f^{ij}\f_{ij } ] \tilde v-2 f^{ij } \tilde v_i \f_j \end{aligned}\ ] ] the parabolic inequality @xmath345 f^{ij } h^k_i h_{kj }
\tilde v \f + c[\e^{-1}+\abs\lam \m e^{\lam u}]f^{ij } g_{ij } \tilde v^3 \f\\[\cma ] & \hp{\le } + [ \frac1{1 - 2\e}-1 ] f^{ij } u_i u_j \m^2\lam^2 e^{2\lam u } \tilde v \f \\[\cma ] & \hp{\le}-f^{ij } u_i u_j \m\lam^2 e^{\lam u } \tilde v \f \\[\cma ] & \hp{\le } + f [ -\h_{\al\bet}\n^\al \n^\bet+e^{-\psi } \m \lam e^{\lam u } \tilde v^2 ] \f\\[\cma ] & \hp{\le } + c\msp[2 ] \nnorm{f_\bet } \tilde v^2 \f + f_{\n^\bet } x^\bet_l h^{kl } u_k e^\psi \f , \end{aligned}\ ] ] where we have chosen the same function @xmath319 in resp . .
setting @xmath346 and using , the assumption , which can be rewritten as @xmath347 as well as the assumptions , , and , and observing , furthermore , that in view of the concavity and homogeneity of @xmath0 @xmath348 we conclude @xmath349 & \hp{\le } + \frac2{1 - 2\e } f^{ij } u_i u_j \m^2\lam^2 e^{\lam u } \tilde v \f -f^{ij } u_i u_j \m\lam^2 e^{\lam u } \tilde v \f \\[\cma ] & \hp{\le}+ c \msp c_1 \m \abs\lam e^{\lam u } \tilde v^2 \f + c\msp c_2 \tilde
v^3 \f + c\msp c^2_3 \e_0^{-1 } e^{\lam u}\tilde v^3 \f , \end{aligned}\ ] ] where @xmath350 is chosen so large that @xmath351 choosing , furthermore , @xmath352 we see that the terms involving @xmath353 add up to a dominating negative quantity that can be estimated from above by @xmath354 in view of .
@xmath355 is of the order @xmath356 for large @xmath330 , hence , the parabolic maximum principle yields a uniform estimate for @xmath357 if @xmath350 is chosen large enough .
we want to prove that the principal curvatures of the flow hypersurfaces are uniformly bounded .
let @xmath245 , @xmath358 , be solutions of the evolution problem with @xmath359 , @xmath360 , and @xmath361 strictly positive , @xmath362 then , the principal curvatures of the flow hypersurfaces are uniformly bounded provided the @xmath245 are uniformly space - like , i.e. uniform @xmath51- estimates are valid . as we have already mentioned in , we know that @xmath363 during the evolution , thus , it is sufficient to estimate the principal curvatures from above .
let @xmath321 be defined by @xmath364 we claim that @xmath321 is uniformly bounded .
let @xmath365 , and @xmath366 , with @xmath367 , be a point in @xmath368 such that @xmath369 we then introduce a riemannian normal coordinate system @xmath122 at @xmath370 such that at @xmath371 we have @xmath372 let @xmath373 be the contravariant vector field defined by @xmath374 and set @xmath375 @xmath376 is well defined in neighbourhood of @xmath377 , and @xmath376 assumes its maximum at @xmath377 .
moreover , at @xmath377 we have @xmath378 and the spatial derivatives do also coincide ; in short , at @xmath377 @xmath376 satisfies the same differential equation as @xmath379 . for the sake of greater clarity , let us therefore treat @xmath379 like a scalar and pretend that @xmath380 .
at @xmath377 we have @xmath381 , and , in view of the maximum principle , we deduce from @xmath382 & \hp{\le}+c(1+\abs a^2 ) ( 1+f+\nnorm{df}+\nnorm{d^2f } ) , \end{aligned}\ ] ] where we used the concavity of @xmath0 , the codazzi equations , , and where @xmath383 thus , @xmath321 is uniformly bounded in view of .
we shall show that the solution of the evolution problem exists for all time , and that it converges to a stationary solution .
the solutions @xmath278 of the evolution problem with @xmath243 , and @xmath359 exist for all time and converge to a stationary solution provided @xmath361 satisfies the conditions , , and .
let us look at the scalar version of the flow as in @xmath384 this is a scalar parabolic differential equation defined on the cylinder @xmath385 with initial value @xmath386 .
in view of the a priori estimates , which we have established in the preceding sections , we know that @xmath387 and @xmath388 independent of @xmath389 due to the definition of the class @xmath219 .
thus , we can apply the known regularity results , see e.g. , where even more general operators are considered , to conclude that uniform @xmath28-estimates are valid , leading further to uniform @xmath25-estimates due to the regularity results for linear operators .
therefore , the maximal time interval is unbounded , i.e. @xmath390 .
now , integrating with respect to @xmath389 , and observing that the right - hand side is non - positive , yields @xmath391 i.e. , @xmath392 hence , for any @xmath393 there is a sequence @xmath394 such that @xmath395 . on the other hand , @xmath396 is monotone decreasing and therefore @xmath397 exists and
is of class @xmath398 in view of the a priori estimates .
we , finally , conclude that @xmath399 is a stationary solution , and that @xmath400 an immediate consequence of the results we have proved so far cf . especially and is the following theorem which is of independent interest .
let @xmath57 be a concave curvature function vanishing on @xmath223 and homogeneous of degree 1 .
let @xmath27 of class @xmath28 satisfy the conditions , , and , and suppose that the boundary components @xmath33 act as barriers for @xmath34 , then , there exists an admissible hypersurface @xmath401 , @xmath402 , solving @xmath403 for small @xmath91 .
we want to solve the equation @xmath404 where @xmath11 satisfies the conditions of .
the curvature function @xmath405 is concave and the elliptic regularization @xmath55 of class @xmath219 , cf . and .
thus , we would like to apply the preceding existence result to find hypersurfaces @xmath406 such that @xmath407 but , unfortunately , the derivatives @xmath408 grow quadratically in @xmath409 contrary to the assumption in . therefore , we define a smooth cut - off function @xmath410_+)$ ] , @xmath411 , where @xmath412 is to be determined later , by @xmath413 such that @xmath414 and consider the problem @xmath415 where for a space - like hypersurface @xmath206 with past directed normal vector @xmath128 as in , we set @xmath416 then , @xmath417 so that the assumptions in are certainly satisfied
. the constant @xmath418 should be so large that @xmath419 in case of the barriers @xmath33 , @xmath263 . if we now start with the evolution equation @xmath420 then , the @xmath33 are barriers for @xmath421 for small @xmath182 , cf . and we conclude the flow hypersurfaces @xmath422 stay in @xmath423 during the evolution if @xmath182 is small @xmath424 .
when we consider the elliptic regularizations @xmath55 , we would like to generalize the meaning of _ admissible _ hypersurface by calling a hypersurface admissible if the tensor @xmath425 is admissible , i.e. if its eigenvalues belong to @xmath36 .
next , let us consider the evolution equations for @xmath330 and @xmath426 which look slightly different : in the term involving @xmath427 has to be replaced by @xmath428 x^\al_k g^{ik}\h_\al.\ ] ] but in view of , the additional terms do not cause any new problems in the proof of , and hence , the uniform @xmath51- estimates are still valid for the modified evolution problem , where the estimates depend on @xmath79 . the @xmath52- estimates in remain valid ,
too , since the second derivatives of @xmath429 , @xmath430 , that occur on the right - hand side of , can be expressed as we only consider the covariant form @xmath431 , no summation over @xmath78 @xmath432 & - \tilde f_\al\n^\al h_{ii}- \tilde f _ { \tilde \n^\al \tilde \n^\bet } \tilde \n^\al_i \tilde \n^\bet_i -\tilde f_{\tilde\n^\al}\tilde\n^\al_{i;i } , \end{aligned}\ ] ] where @xmath433 @xmath434 & \hp{= } + \ddot\tht \tilde v_i \tilde v_i \tilde v^{-1}\n^\al - 2\dot\tht \tilde v_i \tilde v_i \tilde v^{-2}\n^\al+\dot\tht \tilde v_{ii}\tilde v^{-1}\nu^\al\\[\cma ] & \hp{=}+ 2\tht \tilde v^{-3 } \tilde v_i \tilde v_i\n^\al-\tht \tilde v^{-2 } \tilde v_{ii } \n^\al , \end{aligned}\ ] ] and @xmath435 hence , the result of is still valid since no additional bad terms occur in inequality as one easily checks , and since , furthermore , we also have @xmath436 during the evolution , for the modified version of now has the form @xmath437 & = - f^{ij}_\e h_{ik}h_j^k ( f_\e- \tilde f ) - \tilde f_\al\n^\al ( f_\e- \tilde f)\\[\cma ] & \hp{=\ ; } - \tilde f _ { \tilde \n^\ga}\n^\ga [ \dot\tht \tilde v^{-1}-\tht \tilde v^{-2 } ] \h_{\al\bet } \n^\al \n^\bet ( f_\e- \tilde f)\\[\cma ] & \hp{=\ ; } -[\dot\tht \tilde v^{-1}-\tht \tilde v^{-2 } ] \tilde f _ { \tilde \n^\bet}\n^\bet \h_\al x^\al_i ( f_\e- \tilde f)_jg^{ij}\\[\cma ] & \hp{+\ ; } -\tht \tilde v^{-1 } \tilde f _ { \tilde \n^\al } x^\al_i ( f_\e - \tilde f)_j g^{ij}\\[\cma ] & \hp{=\ ; } - f^{ij}_\e\riema \al\bet\ga\de\n^\al x_i^\bet \n^\ga x_j^\de ( f_\e-\tilde f ) . \end{aligned}\ ] ] here , we used the relation @xmath438 which follows immediately from , together with and .
the conclusions of are therefore applicable leading to a solution of equation .
consider the solutions @xmath439 of equation , which at the moment not only depend on @xmath182 but also on @xmath79 , the parameter of the cut - off function @xmath440 , cf . .
we shall prove that the hypersurfaces @xmath218 are uniformly space - like independent of @xmath182 and @xmath79 , or , equivalently , that there exists a constant @xmath441 such that @xmath442 where the parameter @xmath182 is supposed to be small and @xmath79 to be large , so that the barrier condition is satisfied .
let @xmath443 be a solution of , then , the estimate is valid uniformly in @xmath182 and @xmath79 .
hence , @xmath439 is a solution of equation , if we choose @xmath444 . for arbitrary but fixed values of @xmath182 and @xmath79 ,
let us introduce the notation @xmath445 for @xmath55 , where from now on through the rest of the article @xmath446 and where @xmath27 satisfies @xmath447 \nnorm{f_\bet(x,\n)}&\le c_2(1 + \nnorm\n^2),\lae{8.4}\\ \intertext{and } \nnorm{f_{\n^\bet}(x,\n)}&\le c_3 ( 1+\nnorm\n),\lae{8.5}\end{aligned}\ ] ] for all @xmath30 and all past directed time - like vectors @xmath31 .
thus , @xmath0 is homogeneous of degree 2 , and we recall that @xmath448 and @xmath449 where @xmath450 is evaluated at @xmath451 and @xmath452 at @xmath453 .
we also drop the index @xmath182 , writing @xmath123 for @xmath443 and @xmath10 for @xmath218 , i.e. @xmath10 solves the equation @xmath454 the @xmath51- estimate will follow the arguments in the proof of , where at one point we shall introduce an additional observation especially suitable for the curvature function @xmath18 . the former parabolic equations and inequalities , , , and
can now be read as elliptic equations resp .
inequalities by simply assuming that the terms involved are time independent .
though , to be absolutely precise , one has to observe that the present curvature function is homogeneous of degree 2 , which means that , whenever the term @xmath0not derivatives of @xmath0occurs explicitly in the equations or inequalities just mentioned , it has to be replaced by @xmath455 because it was obtained as a result of euler s formula for homogeneous functions of degree @xmath456 @xmath457 we mention it as a matter of fact only , since it does nt affect the estimates at all .
however , we have to be aware that @xmath11 now depends on @xmath458 instead of @xmath128 , i.e. the elliptic version of inequality now takes the form @xmath459 \tilde f^{ij } h^k_i h_{kj } \tilde v \f + c[\de^{-1}+\abs\lam
\m e^{\lam u } ] \tilde f^{ij } g_{ij } \tilde v^3 \f\\[\cma ] & \hp{\le } + [ \frac1{1 - 2\de}-1 ] \tilde f^{ij } u_i u_j \m^2\lam^2 e^{2\lam u } \tilde v \f \\[\cma ] & \hp{\le}-\tilde f^{ij } u_i u_j \m\lam^2 e^{\lam u } \tilde v \f \\[\cma ] & \hp{\le } + 2f [ -\h_{\al\bet}\n^\al \n^\bet+e^{-\psi } \m \lam e^{\lam u } \tilde v^2 ] \f + c\msp[2 ] \nnorm{f_\bet } \tilde v^2 \f \\[\cma ] & \hp{\le } + f _ { \tilde \n^\bet}[\tht \tilde v ^{-1}\n^\bet_i + \dot\tht \tilde v _
i \tilde v^{-1 } \n^\bet -\tht \tilde v^{-2 } \tilde v_i \n^\bet ] u^i e^\psi \f . \end{aligned}\ ] ] here , we used the notation @xmath460 for the small parameter in the schwarz inequality instead of @xmath182 , which has a different meaning in the present context , @xmath357 is defined as in , where the parameters @xmath461 should satisfy the conditions in , and @xmath123 is supposed to be less than @xmath462 .
we claim that @xmath357 is uniformly bounded provided @xmath463 and @xmath464 are chosen appropriately . following the arguments in
, we shall use the maximum principle and consider a point @xmath465 , where @xmath466 as before , we choose @xmath467 .
but the further conclusions are no longer valid , since we have a really bad term on the right - hand side of that is of the order @xmath468 due to the assumption .
the only possible good term which can balance it , is @xmath469 \tilde f^{ij } h^k_i h_{kj } \tilde v \f.\ ] ] to exploit this term we use the fact that @xmath470 , or , equivalently @xmath471 & = e^\psi h^k_iu_k -\h_{\al\bet}\n^\al x^\bet_i , \end{aligned}\ ] ] where the second equation follows from and the definition of the covariant vectorfield @xmath472 .
next , we choose a coordinate system @xmath122 such that in the critical point @xmath473 and the labelling of the principal curvatures corresponds to @xmath474 then , we deduce from @xmath475 assume that @xmath476 , and let @xmath477 be an index such that @xmath478 setting @xmath479 and assuming without loss of generality that @xmath480 in @xmath247 we infer from @xmath481 & \le \m\lam e^{\lam u } \tilde v u_ie^i + c \tilde
v^2 , \end{aligned}\ ] ] and we deduce further in view of and that @xmath482 \tilde v e^{-\psi } \le \frac12\m\lam e^{\lam u } \tilde v e^{-\psi } , \end{aligned}\ ] ] if @xmath350 is sufficiently large , i.e. @xmath483 is and of the same order as @xmath330 .
the weingarten equation and yield @xmath484^{\frac12},\ ] ] and therefore , we infer from @xmath485 in critical points of @xmath357 , and hence , that in those points , the term involving @xmath486 on the right - hand side of inequality can be estimated from above by @xmath487 u^i e^\psi \f}\le c \m\abs\lam e^{\lam u } \tilde v^4\f.\ ] ] next , let us estimate the crucial term in .
using , the particular coordinate system , as well as the inequalities , together with the fact that @xmath483 is negative , we conclude @xmath488 & \le -\sum_{i=1}^{i_0 } f^i_i \ka^2_{i_0 } , \end{aligned}\ ] ] where we recall that the argument of @xmath489 is the @xmath121- tupel with components @xmath490 and observe that in the present coordinate system @xmath491 let @xmath492 , then , @xmath493 is concave , and therefore , we have in view of @xmath494 cf . , or equivalently , @xmath495 hence , we conclude @xmath496 & = -\frac1n(n-1)[h+\e n h]\le -\frac{n-1}n h , \end{aligned}\ ] ] where we also used and . combining , , , and the estimate , we deduce further @xmath497 & \le -\frac{n-1}{2n } c_n \abs{\ka_{i_0}}^3-\frac{n-1}{2n } h \ka_{i_0}^2\\[\cma ] & \le -a_0 \m^3\abs\lam^3 e^{3\lam u } \tilde v^3-a_1 h \m^2\lam^2 e^{2\lam u } \tilde v^2 \end{aligned}\ ] ] with some positive constants @xmath498 and @xmath499 . inserting this estimate , and the estimate in in the elliptic inequality , with @xmath467 , we finally obtain @xmath500 & \hp{\le } + \frac2{1 - 2\de}\msp[2]\tilde f^{ij } u_i u_j \m^2\lam^2 e^{\lam u } \tilde v \f + c [ 1+\abs\lam \m ] h e^{\lam u } \tilde v^3\f\\[\cma ] & \hp{\le}-\tilde f^{ij } u_i u_j \m\lam^2 e^{\lam u } \tilde v \f + c[c_2+c_3 \m\abs\lam e^{\lam u } ] \tilde v^4 \f\\[\cma ] & \hp{\le } + 2f [ c + e^{-\psi } \m \lam e^{\lam u } ] \tilde v^2\f .
\end{aligned}\ ] ] choosing , now , @xmath501 and @xmath350 large , the right - hand side of the preceding inequality is negative , contradicting the maximum principle , i.e. the maximum of @xmath357 can not occur at point where @xmath502 .
thus , the desired uniform estimate for @xmath357 and , hence , @xmath330 is proved .
let us close this section with an interesting observation that is an immediate consequence of the preceding proof , we have especially and the first line of inequality in mind , let @xmath503 be a positive symmetric curvature function such that the partial derivatives @xmath504 are positive and @xmath505 is concave .
suppose @xmath0 is evaluated at a point @xmath35 , and assume that @xmath483 is a component that is either negative or the smallest component of that particular @xmath121- tupel , then @xmath506 \ka^2_{i_0}.\ ] ]
we want to prove uniform @xmath52- estimates for admissible solutions @xmath10 of @xmath507 where we use the notations and conventions of the preceding section .
the starting point is an elliptic equation for the second fundamental form .
the tensor @xmath426 satisfies the elliptic equation @xmath508 the elliptic equation can be immediately derived from the corresponding parabolic equation by dropping the time - derivative , replacing @xmath0 by @xmath113 , and observing that the present curvature function is homogeneous of degree 2 , cf . .
contracting over the indices @xmath509 in we obtain a differential equation for @xmath84 @xmath510 & = - \tilde f^{kl}h_{rk}h_l^rh + 2f \abs a^2\\[\cma ] & \hp{+}- f_{\al\bet } x_i^\al x_k^\bet
g^{ki}- f_\al\n^\al h-2f_{\al\n^\bet}x^\al_i x^\bet_kh^{ki}\\[\cma ] & \hp{= } -f_{\n^\al\n^\bet}x^\al_lx^\bet_kh^k_ih^{li}-f_{\n^\bet}(h^k x^\bet_k + \abs
a^2\n^\bet)\\[\cma ] & \hp{=}-\bar r_{\al\bet}\n^\al x^\bet_kg^{kl}x^\ga_l f_{\n^\ga } + \tilde f^{kl , rs}h_{kl;i}h_{rs;}^{\hphantom{rs;}i}\\[\cma ] & \hp{=}+2 \tilde f^{kl}\riema \al\bet\ga\de x_m^\al x_i ^\bet x_k^\ga x_r^\de h_l^m g^{ri } - 2\tilde f^{kl } \riema \al\bet\ga\de x_m^\al x_k ^\bet x_i^\ga x_l ^\de h^{mi}\\[\cma ] & \hp{=}- \tilde f^{kl}\riema \al\bet\ga\de\n^\al x_k^\bet\n^\ga x_l^\de h+ 2f \bar r_{\al\bet}\n^\al\n^\bet\\[\cma ] & \hp{=}+ \tilde f^{kl}\bar r_{\al\bet\ga\de;\e}\{\n^\al x_k^\bet x_l^\ga x_i^\de x_m^\e g^{mi}+\n^\al x_i^\bet x_k^\ga x_m^\de x_l^\e g^{mi}\ } , \end{aligned}\ ] ] where we also used the symmetry properties of the riemann curvature tensor and the codazzi equations at one point . next ,
let us improve the estimate in .
let @xmath206 be an admissible solution of , then the principal curvatures of @xmath10 satisfy the estimate @xmath511 where the constant depends on @xmath512 , the constant @xmath513 in , and on known estimates of the @xmath514 and @xmath51- norm of @xmath123 .
we argue as in the proof of and define @xmath515 let @xmath465 be a point , where @xmath321 achieves its maximum , and assume without loss of generality that , after having introduced normal riemannian coordinates around @xmath247 , we may write @xmath516 , cf . the corresponding arguments in the proof of . applying the maximum principle in @xmath247 , we deduce from the following inequality @xmath517 & \hp{\le\ ; } + c(1+\nnorm{df}+\nnorm{d^2f})(1+\abs a^2)+c ( \tilde f^{ij}g_{ij}+f ) , \end{aligned}\ ] ] where we also used , the weingarten and codazzi equations , and the fact that the pair @xmath53 stays in a compact subset of @xmath518 , where @xmath519 stands for the set of past directed time - like vectorfields in @xmath48 . furthermore , we know that @xmath520 and @xmath521 & \le c\msp[2 ] c_1^{-1 } ( 1+\abs a^2 ) , \end{aligned}\ ] ] since @xmath522 is concave , cf . . thus , we conclude that in @xmath247 the following inequality is valid @xmath523 with a known constant @xmath343 , and the lemma is proved .
the estimate will play an important role in the final a priori estimate .
let @xmath18 , @xmath524 a riemannian manifold with metric @xmath250 , @xmath451 a symmetric tensor field on @xmath10 the eigenvalues of which belong to @xmath36 , and @xmath212 an arbitrary point .
choose local coordinates around @xmath294 such that the relations and are satisfied .
then , we have for @xmath525 @xmath526 \sum_{i\ne n}\ka^2_i+2f&\le c f^n_n\ka_n,\lae{9.11}\end{aligned}\ ] ] and @xmath527 with @xmath528 , @xmath0 is evaluated at @xmath451 , and where we point out that the summation convention is not used . throughout the proof we shall use the ambivalent meaning of @xmath0 as a function depending on @xmath71 or on @xmath451 switching freely from one viewpoint to the other .
\(i ) from the definition of @xmath0 @xmath87 and we conclude @xmath529 & = \abs{f^j_j}^2 + 2f^j_j\ka_j-\sum_{i\ne j}\ka^2_i , \end{aligned}\ ] ] which proves .
\(ii ) if @xmath530 , and thus @xmath531 the largest eigenvalue , then , we derive from @xmath532 and follows at once from .
\(iii ) a simple algebraic transformation yields @xmath533 & = \frac{f_{;j}}{hf^j_j}(\sum_{k\ne j}\ka_k+\ka_i ) , \end{aligned}\ ] ] and hence , @xmath534 \frac{\abs{f_{;j}}^2}{h^2\abs{f^j_j}^2}\sum_{i\ne j}\ka^2_i . \end{aligned}\ ] ] we , now , treat the cases @xmath530 and @xmath535 separately . if @xmath530 , we apply and and conclude @xmath536 \frac{\abs{f_{;j}}^2}{hf^j_j}\le c\msp[2 ] \frac{\abs{f_{;j}}^2}{f}. \end{aligned}\ ] ] if @xmath535 , we deduce from @xmath537 and deduce further @xmath538 where apparently we only had to worry about the case @xmath539 .
thus , the right - hand side of is estimated from above by @xmath540\frac{\abs{f_{;j}}^2}{h^2},\ ] ] which in turn is less than @xmath540\frac{\abs{f_{;j}}^2}f.\ ] ] let @xmath10 be an admissible solution of and @xmath212 arbitrary .
choose local coordinates around @xmath294 such that the relations and are valid .
then , for any @xmath525 , the following inequality is valid in @xmath294 @xmath541 where we use the notation @xmath542 , and do not apply the summation convention .
let us recall the relation , which we can also express in the form @xmath543 where @xmath544 consider each summand in separately .
we have @xmath545 & \le \frac{f_j^2 } { \widetilde h ^2 \abs { f^j_j}^2}\absb{\sum_{k\ne j } \tilde \ka_k+ \tilde \ka_i}^2 , \end{aligned}\ ] ] in view of , i.e. we are exactly in the same situation as in the proof of after the equation with the following modifications : we replace @xmath546 and @xmath84 by @xmath547 resp .
@xmath548 and observe that @xmath549 .
let @xmath10 be an admissible solution of equation , then , the estimate @xmath550 is valid in every point , where the smallest principal curvature @xmath551 satisfies @xmath552 the proof is a modification of a similar result in or .
it follows immediately from the definition of @xmath0 that @xmath553 and @xmath554 & \hp{=\ ; } + \e[2(n-1 ) + \e(n-1)n ] \norm{dh}^2 .
\end{aligned}\ ] ] in a fixed point @xmath212 introduce normal riemannian coordinates such that the relations and are valid , and define the matrix @xmath555 through @xmath556 we also set @xmath557 then , we conclude from @xmath558 & = \sum_{i , k , l } ( h_{kki } h_{lli}-h_{kli}h_{kli})\\[\cma ] & = \sum_i a^{kl}(h_{kki } h_{lli}- h^2_{kli})\\[\cma ] & = \sum_i a^{kl } h_{kki } h_{lli}- \sum_l\sum_{k , i } a^{kl } h^2_{kli}. \end{aligned}\ ] ] in the last summand let us interchange the roles of @xmath78 and @xmath559 to obtain @xmath560 the codazzi equations yield @xmath561 where @xmath562 is a uniformly bounded tensor in @xmath48 @xmath563 and hence , @xmath564 and @xmath565 & \le - 2\sum_i\sum_k a^{ki } h^2_{kki } - \sum_{[i , j , k ] } h^2_{ijk}\\[\cma ] & \hp{\le\ ; } -2\sum_{i , k , l } a^{ki } h_{kli } c_{kil}-2\sum_{i , k}a^{ki}h_{iik}c_{iki } , \end{aligned}\ ] ] where @xmath566}$ ] means that the summation is carried out over those triples @xmath567 where all three indices are different from each other . the first linear term in the last inequality can be estimated from above by @xmath568 & \le \frac{\de}2 \sum_i\sum_k a^{ki } h^2_{kki}+\frac{\de}2 \sum_{[i , j , k ] } h^2_{ijk } + c\de^{-1 } \end{aligned}\ ] ] for any @xmath176 , and the second term similarly .
thus , we deduce from and @xmath569 & \hp{\le\ ; } + \sum_i a^{kl } h_{kki}h_{lli } + c\de^{-1 } \end{aligned}\ ] ] for any @xmath570 .
next , let us consider the second term on the left - hand side of ; we have @xmath571 & \le \widetilde h^{-1 } \sum_i \tilde f^{ii } ( \sum_kh_{kki})^2 + \e c \norm{dh}^2 , \end{aligned}\ ] ] where @xmath528 , in view of , , and . combining and the preceding estimate ,
we conclude @xmath572 & \le -2(1-\frac\de2 ) \sum_i\sum_k a^{ki } h^2_{kki } + \sum_i a^{kl } h_{kki}h_{lli } + c\de^{-1}\\[\cma ] & \hp{\le\ ; } + \widetilde h^{-1 } \sum_i \tilde f^{ii } ( \sum_kh_{kki})^2 + \e c \norm{dh}^2 . \end{aligned}\ ] ] for each index @xmath78 , let us estimate the corresponding summand separately , i.e. let us look at no summation over @xmath78 @xmath573 where we have divided the terms by @xmath574 . denote by @xmath575 a sum where the index @xmath78 is omitted during the summation , then , can be expressed as @xmath576 & + \frac1{2 \widetilde h } \tilde f^{ii } ( \sum_k h_{kki})^2 . \end{aligned}\ ] ] to replace @xmath577 in the preceding expression we use the chain rule @xmath578 to derive @xmath579 inserting in we obtain , after some simple algebraic manipulations , cf . or , @xmath580h^2_{kki}\\[\cma ] & - { \sum_{k < l}}'\big[\frac { \tilde
f^k_k+ \tilde f^l_l } { \tilde f^i_i}-1-\frac { \tilde f^i_i } { \widetilde h}\big(1-\frac { \tilde f^k_k } { \tilde f^i_i}\big)\big(1-\frac { \tilde f^l_l } { \tilde f^i_i}\big)\big]h_{kki } h_{lli}\\[\cma ] & + { \sum_k}'\big[\frac { f_i } { \tilde f^i_i}+\frac { f_i } { \widetilde h}\big(1-\frac { \tilde f^k_k } { \tilde f^i_i}\big)\big]h_{kki } + \frac { \tilde f^i_i}{2 \widetilde h}\big(\frac{f_i } { \tilde f^i_i}\big)^2 . \end{aligned}\ ] ] let us write as the sum of three expressions @xmath581 , where @xmath582h_{kki},\ ] ] @xmath583 and @xmath584h^2_{kki}\\[\cma ] & - { \sum_{k
< l}}'\big[\frac { \tilde f^k_k+ \tilde f^l_l } { \tilde f^i_i}-1-\frac { \tilde f^i_i } { \widetilde h}\big(1-\frac { \tilde f^k_k } { \tilde f^i_i}\big)\big(1-\frac { \tilde f^l_l } { \tilde f^i_i}\big)\big]h_{kki } h_{lli}. \end{aligned}\ ] ] in view of we can estimate @xmath585 from above by @xmath586\de^{-1 } f^{-1 } \abs{f_i}^2.\ ] ] @xmath587 is estimated by @xmath588 because of . finally , we claim that @xmath589 if we choose @xmath590 .
to verify this assertion , let us multiply @xmath591 by @xmath592 to obtain @xmath593h^2_{kki}\\[\cma ] & -{\sum_{k < l}}'[2 \widetilde h ( \tilde f^k_k+ \tilde f^l_l)-2 \widetilde h \tilde f^i_i-2 ( \tilde f^i_i- \tilde f^k_k ) ( \tilde f^i_i- \tilde f^l_l ) ] h_{kki } h_{lli}. \end{aligned}\ ] ] now , we use and replace @xmath594 , @xmath525 , by @xmath595 the expression in is then equal to the sum of two terms @xmath596 , where @xmath597h^2_{kki}\\[\cma ] & -{\sum_{k < l}}'[2 \widetilde h ( f^k_k+ f^l_l)-2 \widetilde h f^i_i-2 ( f^i_i- f^k_k ) ( f^i_i- f^l_l ) ] h_{kki } h_{lli } , \end{aligned}\ ] ] and @xmath598 - 2 \widetilde h \e \ga_\e h{\sum_{k < l } } ' h_{kki}h_{lli}. \end{gathered}\ ] ] from the the binomial formula @xmath599 we infer that @xmath600 , while @xmath601 is non - positive provided we choose @xmath590 and assume @xmath602 cf . or .
but the condition is certainly satisfied in view of . combining , , , and gives , and
thus , the lemma is proved . from
and we conclude let @xmath206 be an admissible solution of equation in @xmath3 .
then , the estimate @xmath603 is valid in every point @xmath212 , where is satisfied .
the constant @xmath343 depends on @xmath3 , @xmath512 , the constant @xmath513 in , and on known estimates of the @xmath514 and @xmath51- norm of @xmath123 . as we already mentioned
we have to assume the existence of a strictly convex function @xmath45 , i.e. @xmath46 satisfies @xmath604 with a positive constant @xmath605 .
we observe that then the restriction @xmath606 of @xmath46 to an admissible solution @xmath43 of satisfies the elliptic inequality @xmath607 & \le -2 \tilde f\chi _ { { \vphantom{}}_\al } \n^\al -c_0 \tilde f^{ij}g_{ij } , \end{aligned}\ ] ] where we used the homogeneity of @xmath113 .
we can now prove uniform @xmath52- estimates .
let @xmath206 be an admissible solution of equation in @xmath3 , where @xmath11 satisfies the estimates , and .
then , the principal curvatures of @xmath10 are uniformly bounded .
let @xmath46 be the strictly convex function and @xmath463 a large positive constant .
we shall prove that @xmath608 is uniformly bounded from above .
let @xmath465 be such that @xmath609 and choose in @xmath247 a local coordinate system satisfying and . applying the maximum principle , we conclude from and @xmath610 & \hp{\le \ ; } + c(1+f+\nnorm{df}+\nnorm{d^2f})(1+h+\norm{d\log h})\\[\cma ] & \hp{\le \ ; } + \tilde f^{ij , kl}h_{ij;r}h_{kl;s } g^{rs}h^{-1 } + \tilde f^{ij } ( \log h)_i ( \log h)_j , \end{aligned}\ ] ] where we also assumed @xmath84 to be larger than @xmath106 .
we now consider two cases .
: suppose that @xmath611 then , we infer from and @xmath612 moreover , the concavity of @xmath522 implies @xmath613 & = f^{-1 } \norm{df}^2 h^{-1}\\[\cma ] & \le c f^{-1}\nnorm{df}^2
\abs a^2h^{-1}\\[\cma ] & \le c f^{-1 } \nnorm{df}^2 h. \end{aligned}\ ] ] furthermore , @xmath470 , or , @xmath614 inserting the last three relations in we obtain @xmath615 where , now , @xmath343 depends on @xmath11 and its derivatives in the ambient space .
hence , @xmath84 , and therefore @xmath357 , are a priori bounded in @xmath247 .
: suppose that @xmath616 then , is applicable , and we infer from and @xmath617 choosing now @xmath463 sufficiently large we obtain an a priori bound for @xmath618 , since @xmath619 thus , @xmath357 , or equivalently @xmath84 , are uniformly bounded .
we can now demonstrate the final step in the proof of .
let @xmath620 be the stationary approximations . in the preceding sections we have proved uniform estimates for @xmath443 up to the order two . since , by assumption
, @xmath11 is strictly positive , the principal curvatures of @xmath218 stay in a compact subset of the cone @xmath36 for small @xmath182 , cf . , and therefore , the operator @xmath445 is uniformly elliptic for those @xmath182 .
taking the square root on both sides of equation without changing the notation , we also know that @xmath113 is concave .
r. p. geroch & g. t. horowitz , _ global structure of space - time _
, general relativity .
an einstein centenary survey ( s. w. hawking & w. israel , eds . ) , cambridge university press , cambridge , 1979 , pp . | the existence of closed hypersurfaces of prescribed scalar curvature in globally hyperbolic lorentzian manifolds is proved provided there are barriers . |
the @xmath4 and @xmath5 damped ly@xmath1 ( dla ) absorption - line systems along the sight - line toward the quasar oi@xmath0363 ( 0738 + 313 ) have been the subject of several recent papers by members of our group .
rao & turnshek ( 1998 , hereafter rt98 ) announced the discovery of these systems during the course of an hst - fos survey for absorption in qso mgii systems . in this paper
they discussed the results of a preliminary optical imaging search for galactic light from these two absorbers , and they concluded that the galactic counterparts are definitely not luminous spirals .
lane et al .
( 1998a , b ) and lane , briggs & smette ( 2000a ) presented the results of 21 cm observations along the sight - line toward oi@xmath0363 .
they found that both of the damped systems exhibit 21 cm absorption .
however , an initial search for 21 cm emission from the lower - redshift system failed to detect a signal ( lane et al .
2000b ) .
the existence of these two systems is noteworthy because they currently represent the two lowest - redshift classical quasar absorption - line systems which are confirmed , i.e. , actual uv spectroscopic observations of their profiles exist and analyses of the spectra demonstrate that the systems neutral hydrogen column densities are above the classical survey limit of @xmath2 atoms @xmath3 . in several other reported cases of low - redshift systems , either confirming uv spectroscopy is lacking ( miller , knezek & bregman 1999 ) , the h1 column density is somewhat below the classical limit ( lanzetta et al .
1997 ) , or both .
thus , although detection of foreground galaxies near a bright background qso remains difficult , the oi@xmath0363 sight - line offers a most ideal opportunity for studying any emission from stellar and gaseous components of objects responsible for classical absorbers .
we refer to such luminous objects as galaxies . here
we report on the results of new observations of galaxies along the sight - line toward oi@xmath0363 .
the data consist of optical and infrared imaging and also optical spectroscopy in seeing conditions sometimes as good as @xmath19 arcsec . throughout the analysis we assume @xmath20 km s@xmath9
mpc@xmath9 , @xmath10 , and @xmath11 ; we also adopt the naming convention used in rt98 . in
2 we start by summarizing the results from our earlier studies . in
3 the new observations and analyses of these data are presented . in brief , the galaxy associated with the @xmath21 absorber appears to be an extended low surface brightness dwarf galaxy at impact parameter @xmath22 kpc , with rich morphological structure and a k - band luminosity of @xmath23 , where the measured light is an upper limit depending on the amount of contamination from the quasar host galaxy at redshift @xmath24 ( @xmath25 assumes no host contribution ) .
the galaxy associated with the @xmath26 absorber is considerably different in character ; it is an early - type dwarf galaxy which is relatively compact at an impact parameter of @xmath17 kpc , with a k - band luminosity of @xmath27 . in 4 we discuss the implications of these findings .
rt98 presented the hst - fos uv spectrum of oi@xmath0363 ( see their figures 1 - 3 ) which led to the identification of the two absorbing systems at @xmath4 and @xmath5 with h1 column densities of @xmath28 atoms @xmath3 and @xmath29 atoms @xmath3 , respectively .
rt98 also presented a 45 min r - band wiyn image of the oi@xmath0363 field in seeing of 0.55 arcsec ( see their figure 4 ) .
analysis of this image resulted in a list of 15 resolved objects and 11 point sources brighter than r = 24 mag within 40 arcsec of the quasar sight - line ( see their table 1 ) .
the closest resolved object was found to be a galaxy ( named g1 , r = 20.8 mag ) 5.7 arcsec to the south - southeast of the quasar and the closest unresolved object is presumably a star ( named s1 , r = 20.8 mag ) 2.5 arcsec to the northeast of the quasar .
fainter objects were visible in the r - band wiyn image but they were not tabulated since such a list would have suffered from incompleteness . however , the existence of a marginally significant fainter extended feature a few arcsec to the west - southwest of the quasar sight - line was noted ( marked with a cross on figure 4 of rt98 ) .
it appears to be the brightest part of the `` fuzz '' near the quasar , and was reported to have an r - band surface brightness of @xmath30 mag arcsec@xmath31 .
notably , the brightest galaxy in the field is a luminous spiral ( named g11 , r = 17.1 mag ) @xmath32 arcsec to the east of the quasar with a redshift of @xmath33 . a similarly luminous spiral at a somewhat larger redshift of @xmath6 , i.e.
, the redshift of the lowest redshift absorber , could have easily been detected in this field if it was present , but the observations ruled this out .
we note that there is an absorption line at @xmath34 in the original fos low - resolution spectrum of the quasar , which can be identified as ly@xmath1 with a rest equivalent width of w@xmath35 and a redshift of @xmath36 .
the existence of a ly@xmath1 line at @xmath37 is consistent with the results of guillemin & bergeron ( 1996 ) . in any case , due to their close proximity to the quasar sight - line in comparison to other objects in the field
, the galaxy labeled g1 and any object that might be associated with the fuzz are of most significant interest ; the results from our new observations ( 3 ) indeed suggest that these luminous objects can be identified with the two galaxies .
both of the absorbers along this sight - line have been shown to have associated h1 21 cm absorption ( lane et al .
1998a , b and 2000a ; chengalur & kanekar 1999 ) .
the @xmath5 system shows only a single narrow 21 cm absorption feature , while the @xmath4 system has complex velocity structure .
the spectrum from this lower redshift system is best fit by multiple gaussians : two narrow cold components with thermal kinetic widths of t@xmath38 and @xmath39 k , and a third broad component with t@xmath40 k ( lane et al .
this suggests the presence of two temperature phases for the neutral medium , similar to those found in the galaxy . for this lower redshift system ,
excluding the velocity interval where absorption is detected ( a region covering @xmath41 km s@xmath9 ) , an upper limit has been placed on the h1 gas mass .
sensitive wsrt 21 cm emission measurements of the @xmath4 absorber place the upper limit at @xmath42h@xmath43 m@xmath44 for an assumed velocity spread of 100 km s@xmath9 ( lane et al .
this requires the h1 mass of the absorber to be somewhat less than that of a normal spiral galaxy . in principal , since the velocity spread would be lower for a more face - on spiral , the formal h1 mass limit for this special case could be even lower .
however , if the 21 cm emission were confined to a very narrow velocity interval , one might then be concerned that a significant amount of the 21 cm emission were missed because it fell in the same velocity intervals as the absorption .
new observations along the sight - line toward oi@xmath0363 include optical and infrared imaging and optical spectroscopy .
based on our identifications , here we exclusively consider analysis of the data pertaining to galaxy g1 and the fuzz near the quasar .
g1 is shown to be a dwarf galaxy at redshift @xmath45 , consistent with the redshift of the @xmath5 absorber .
we argue that at least part of the fuzz is associated with the @xmath4 absorber , leading to the interpretation that the @xmath6 galaxy is an extended low surface brightness dwarf galaxy with rich morphological structure .
optical bri imaging data of the oi@xmath0363 field were obtained for us by the wiyn queue team using a @xmath46 tektronics stis thinned , frontside illuminated ccd ( 0.195 arcsec per pixel ) on the 3.5-m wiyn telescope on kitt peak .
in addition , we obtained supplemental u - band imaging data using the @xmath47 templeton thinned , backside illuminated ccd ( 0.28 arcsec per pixel ) on the mdm observatory 2.4-m hiltner telescope on kitt peak . the seeing ranged between @xmath48 arcsec .
the journal of observations for these optical images is presented in table 1 .
these data were processed using normal methods .
observations of standard stars and field stars in the quasar frame allowed us to obtain calibrated magnitudes on the johnson ( u , b ) and cousins ( r , i ) system .
we note that the r calibration is consistent with the calibration used in rt98 which was originally based on the calibration of drinkwater , webster , & thomas ( 1993 ) .
however , the four color images are not shown here because they do not reveal any additional qualitative information beyond what can be learned from inspection of the r - band wiyn image that was published by rt98 .
infrared imaging observations of the oi@xmath0363 field in jhk were obtained on the 3.0-m nasa irtf on mauna kea using nsfcam , a 256@xmath49256 insb detector array ( 0.30 arcsec per pixel ) .
the journal of observations for these infrared images is also presented in table 1 .
the observations were obtained during the course of three different observing runs in various observing conditions .
the conditions were not always photometric and there were variations in the infrared sky brightness ; the seeing ranged between @xmath50 arcsec . using recommended procedures ( see the irtf.ifa.hawaii.edu/facility/nsfcam/nsfcam.html web site ) , the observations were obtained using several different short - exposure dither patterns , flatfielded with sky frames that were obtained from the dithered object frames , and then shifted before addition to obtain a `` final '' image . since several different standard stars were used and the data were collected on a variety of dates , this provided multiple opportunities to check the photometric calibrations . panels ( a ) and ( b ) of figure 1 show one of the k band images ( 60 min ) of the oi@xmath0363 field and its associated isophotal plot .
this was one of the best k - band images in the sense that it had good seeing ( @xmath51 0.75 arcsec ) and a low k - band sky brightness , allowing very low surface brightness features to be revealed . for display , this image was smoothed with a gaussian ( fwhm = 0.75 arcsec ) and the unresolved nuclear psf was then subtracted ( see 3.4.2 for discussion of this process ) . due to poorer seeing and/or higher background sky levels ,
the remaining k - band images provided little additional information , with the exception of the checks of the photometric calibration .
optical spectroscopic observations of objects in the oi@xmath0363 field were obtained using the noao 4.0-m mayall telescope on kitt peak .
the aim of the spectroscopy was to identify or exclude objects near the redshifts of @xmath15 and @xmath6 .
we will only present the spectrum of g1 here since the fuzz near the quasar was too faint to obtain a reliable spectrum and the other spectra we obtained did not result in identifications for either of these two redshifts .
spectra of objects in this field were obtained with the cryocam in long - slit and multi - slit modes .
since g1 is both faint and close to the bright qso , obtaining its spectrum was difficult .
we found that the long slits produced stronger object signals that were easier to measure relative to sky than the short multi - slit masks .
the final spectrum of g1 is a combination of two long - slit spectra obtained separately with the grism 650 ( 400 l mm@xmath9 covering @xmath52 ) and the grism 770 ( 300 l mm@xmath9 covering @xmath53 ) .
the bluer spectrum was obtained in 120 minutes on the night of 29 april 1998 with a 1.7 arcsec wide slit at a resolution of @xmath54 ( 3.2 per pixel ) ; the redder spectrum was obtained in 90 minutes on the night of 14 feb 1999 with a 1.0 arcsec wide slit at a resolution of @xmath55 (4.3 per pixel ) .
the slit width used was matched to the seeing conditions as closely as possible .
wavelength calibration of the spectra was accomplished with the observation of an internal henear lamp .
the g1 spectra were extracted from the flatfielded loral ( ford ) ccd images after removal of cosmic rays using an optimal extraction algorithm .
the spectra were initially flux - calibrated using standard star observations , adjusted to the same flux level in the overlapping region and combined by variance weighting .
the observations produced a useful spectrum of g1 over the wavelength interval @xmath56 . the @xmath57 spectrum , rebinned to @xmath51 12 pixels and normalized in flux at 6000 , is shown in figure 2 .
the downturn in the flux beyond 8000 is possibly due to poor calibration at the red end of the grism 770 spectrum where the spectrograph becomes more defocussed . the irtf k - band image ( figure 1 ) shows g1 ( @xmath58 arcsec southeast of the quasar ) and reveals the presence of luminous features in the fuzz surrounding the quasar that were not very significant in the rt98 wiyn r - band image .
the fuzz is seen to consist of features of low , but not uniform , surface brightness .
one gets the impression that , within the fuzz , the elongated feature to the east - southeast of the quasar could be part of a ( spiral ) `` arm '' since it is approximately concentric with respect to the somewhat off - centered light surrounding the quasar ; a similar arm is not apparent on the opposite side of the quasar . to the west - southwest there is a feature elongated from east to west that appears like a `` jet '' since it runs nearly perpendicular to the otherwise concentric - looking extended light .
this jet - like feature appears to have structure .
it is coincident with the cross marked on the original rt98 wiyn r - band image , but it does not emanate from the quasar nucleus . in the discussion below we will use the terms `` arm '' and `` jet '' to describe the appearance of these features , but
there is certainly no _ strong _ evidence that these terms describe their actual physical nature .
table 2 lists the ubrijhk photometric measurements of g1 , the arm , and the jet .
details of how these measurements were made and how they have been used to infer the properties of the two galaxies at @xmath15 and @xmath6 are given below .
since we have a spectrum of g1 , we will consider it first .
_ 3.4.1.1 .
the spectrum _
the spectrum of g1 is shown in figure 2 .
for comparison , the spectrum of an early - type galaxy ( the type e3 elliptical galaxy ngc 4648 , kennicutt 1992a ) , redshifted to @xmath15 and normalized at 6000 , is also plotted .
this template spectrum is offset relative to the spectrum of g1 for clarity .
the spectrum of this early - type galaxy template over our observed wavelength range is dominated by the spectral features of cool giants , i.e. , the 4000 break , the g band at rest wavelength 4304 , the mgi @xmath59 mgh ( mgb ) band at 5175 , the nad lines at 5893 , and a tio band at 6260 .
the positions of these features as they appear in the redshifted spectrum of ngc 4648 are marked .
the telluric b band in this spectrum also appears redshifted , since it was not removed from the original kennicutt ( 1992a ) spectrum .
the unredshifted telluric a and b bands as they appear in the spectrum of g1 are also marked .
we see from its spectrum that g1 is probably best classified as a dwarf elliptical , or possibly e / s0 .
the two most prominent features in the spectrum of g1 that suggest it is at redshift @xmath45 are the 4000 break and the absorption feature that corresponds to the position of the mgi@xmath59mgh band .
the narrow downward spike within this mg band is caused by residuals from subtracting the 6300 night sky line .
the shape of the continuum between 4500 and 8000 matches the e - type galaxy template spectrum remarkably well .
however , given the spectrograph slit size , which corresponds to @xmath51 5 kpc in the blue and @xmath51 3 kpc in the red for @xmath15 , a reasonable interpretation would also be that the light being sampled by the spectrograph originates in the _ bulge _ of an early type spiral galaxy .
template spectra of early type spirals ( s0 and s0/a ) with non - seyfert nuclei ( kennicutt 1992a ) also show the same characteristic continuum shape and spectral features as the e galaxy spectrum shown on figure 2 .
emission lines , in particular h@xmath1 , are weak in the integrated spectra of sa galaxies and are not present in earlier types , so the absence of h@xmath1 and other emission lines in the spectrum of g1 also strongly indicates that g1 is an early type galaxy .
the expected position of h@xmath1 at @xmath15 is marked on figure 2 .
kennicutt ( 1992b ) points out that the equivalent width of [ oii]@xmath603727 emission in galaxies is typically one - third that of h@xmath1 , so the absence of [ oii ] ( which is in the most noisy part of the spectrum ) is consistent with the absence of h@xmath1 ; moreover , when h@xmath1 is weak ( as is the case here ) the equivalent width of [ oiii]@xmath605007 rarely exceeds @xmath61 , so it would also not be expected .
it is interesting that some dwarf es , e / s0s , and s0s are known to have substantial h1 halos ( e.g. lake , schommer & van gorkom 1987 and sadler et al .
2000 ) , extending up to @xmath51 5 times the optical radius .
the h1 column densities in the inner contours easily exceed the 2@xmath62 atoms @xmath3 requirement for producing a classical absorption line . unlike the luminous es and s0s , these early - type dwarf galaxies are h1-rich for their morphological class , with centrally concentrated neutral gas and h1 masses of a few times @xmath63 m@xmath44 . however , these early - type h1-rich dwarfs are evidently rare locally .
the apparent magnitude and spectrum of g1 is reminiscent of some of the faintest galaxies discovered in the hawaii k - band survey of songaila et al .
( 1994 , e.g. , see their object number 265 in field ssa22 ) .
indeed , examination of the statistical results on the hawaii deep survey fields presented by cowie et al .
( 1996 ) indicates that g1 is intrinsically one of the faintest galaxies near redshift @xmath45 in comparison to their sample .
we note that the spectrum of g1 recently obtained by cohen ( 2000 ) is consistent with our interpretation . _
3.4.1.2 . the spectral energy distribution _ since our imaging data were taken with three different telescopes in various conditions and with widely varying limiting magnitudes , an appropriate procedure had to be used to accurately determine g1 s colors .
the worst seeing conditions ( @xmath64 arcsec ) were in the u and h bands .
therefore , all of the other images ( brijk ) were degraded to this resolution by convolution with a gaussian before the photometric measurements were made .
an aperture size of 1.4 arcsec ( @xmath65 kpc at @xmath15 ) was then determined to be a good choice to include most of the galactic light detectable above the sky background in the faintest image and still yield minimal errors .
however , we note that nearly half of g1 s light in the wiyn r - band was not included in the 1.4 arcsec aperture using this method .
extrapolation was then used to derive total magnitudes .
comparison of the wiyn b - band and r - band images , which represents our best two - color data , showed no _ significant _ evidence for a color gradient in g1 .
the estimated errors in g1 s magnitudes ( table 2 ) include both the statistical errors and zero - point calibration errors .
we see that g1 is a neutral - colored dwarf galaxy ( b@xmath66k@xmath67 ) with @xmath68 , @xmath69 and @xmath70 , where an @xmath71 galaxy corresponds to @xmath72 , @xmath73 and @xmath74 ( see loveday 2000 for @xmath75 , marinoni et al .
2000 for @xmath76 , and poggianti 1997 and cowie et al .
1994 for k corrections ) . given the ubrijhk colors of g1 , we point out that further evidence of its redshift comes from a derivation of its photometric redshift . as figure 3 shows
, @xmath77 is the most probable redshift based on fitting 1996 bruzual and charlot galaxy model templates to the colors ( see more discussion below ) .
in particular , it was possible to obtain an excellent match to the spectral energy distribution of g1 using a spectral synthesis code to model a galaxy near redshift @xmath15 .
while we considered a variety of metal abundances , the fits were not very sensitive to metallicity for these colors .
however , slightly higher and lower redshifts could not be ruled out definitively based on the photometry alone ( figure 3 ) and , in the end , the best evidence for the @xmath15 redshift of g1 comes from its slit spectrum ( figure 2 ) and the circumstantial evidence provided by the existence of the @xmath26 absorber in the field .
therefore , while the derivation of photometric redshifts for intrinsically faint galaxies would seem to be a promising method when used with the expectation that there will be a galaxy in the field at the absorption redshift , the technique should be used with caution since most photometric redshift techniques have been tested by comparing photometric and spectroscopic redshifts of intrinsically luminous galaxies , but evidence ( 4 ) suggests that galaxies are often not very luminous . in order to consider how well the observed colors of g1 at @xmath15 match different types of galaxies , and what this might tell us about the present nature of g1 and its progenitor
, we used the publicly available photometric redshift code _ hyperz _
( bolzonella , pello & miralles 2000 ) to produce spectral synthesis models .
first we dereddened the colors of g1 using a galactic extinction law and a@xmath78 ( cardelli , clayton , & mathis 1989 ) .
we then used _ hyperz _ to investigate the range of models and ages that produced acceptable fits to the observed colors of g1 . for our analysis we set the metallicity to 20% of the solar value
the amount of intrinsic visual extinction , a@xmath79 , was allowed to be a free parameter , assuming that the calzetti et al .
( 2000 ) reddening law holds .
the available gissel96 template sets were for an instantaneous single - burst model , plus models with exponentially decreasing star formation rates with e - folding times in gyrs of @xmath80 1 , 2 , 3 , 5 , 15 , 30 , and infinity ( i.e. constant star formation rate ) .
models with @xmath81 are generally taken to correspond to early type galaxies ( e.g. type e or s0 ) .
of these models , the burst models at ages between @xmath82 gyrs , and with a@xmath79 ranging between @xmath83 , respectively , were found to give acceptable fits to the color data .
also , a @xmath84 model with a@xmath85 fits the color data reasonably well at an age of @xmath86 gyrs . a 0.72 gyr old burst model with a@xmath87 mag
gave the best overall fit .
[ we note that for the cardelli et al .
( 1989 ) galactic extinction law , a@xmath88 corresponds to @xmath89 atoms @xmath3 ; while for the calzetti et al .
( 2000 ) extinction law which we used in the _ hyperz _ simulations , a@xmath88 might lie in the range @xmath90 atoms @xmath3 , but a good estimate is hindered by the absence of a metallicity measurement for g1 .
this should be compared to the value observed in the @xmath15 absorber at impact parameter @xmath17 kpc , @xmath91 atoms @xmath3 . ]
figure 4 shows the color data with the models over - plotted .
the best - fitting model is in panel ( a ) and three reasonable alternative models are shown in panels ( b ) , ( c ) , and ( d ) .
these galaxy templates are generally consistent with the observed spectrum of g1 ( figure 2 ) .
a comparison of the model fits shown in figure 4 indicates that far - uv observations could break some of the ubrijhk color degeneracy which makes it hard to distinguish age and extinction in burst models .
thus , lacking far - uv photometry of g1 , we can only infer that the progenitor population originated at a formation epoch between @xmath92 ( corresponding to the 0.36 gyr old burst model ) and @xmath93 ( corresponding to the 3.5 gyr old models ) . for the four models illustrated in figure 4
, we show in figure 5 the corresponding inferred color ( b@xmath66k ) evolutionary history of g1 s progenitor prior to its current state at @xmath15 , including effects due to the k - correction but assuming that no change in intrinsic dust extinction occurred .
as would be expected , in all cases the color evolution suggests that the progenitor of g1 , with b@xmath66k@xmath94 at @xmath15 , would have been considerably bluer close to its formation epoch , i.e. , b@xmath66k@xmath954 at @xmath96 .
therefore , we conclude that g1 might be a faded , low - redshift counterpart originating from the population of moderate - redshift ( @xmath97 ) `` faint blue galaxies '' that have been identified in surveys ( e.g. ellis 1997 and references therein ) .
further details are given in the next section .
_ 3.4.1.3 .
the star formation rate _ an upper limit to the h@xmath1 flux in the spectrum of g1 ( figure 2 ) has been estimated .
this was done by first correcting the continuum shape near 8000 to match that of the figure 2 comparison spectrum .
the absolute flux zero point was then obtained by adjusting this corrected , standard - star - calibrated spectrum by a constant scaling factor to make it agree with the more accurate optical photometric measurements .
measurement of the spectrum using a slit corresponding to a width of @xmath98 kpc near h@xmath1 yields a 4@xmath99 upper limit on the rest frame h@xmath1 flux of @xmath100 ergs @xmath3 s@xmath9 , as measured over an @xmath51 12 interval . at @xmath15
this translates to an h@xmath1 luminosity of @xmath101 ergs s@xmath9 and a corresponding star formation rate @xmath102 m@xmath44 yr@xmath9 ( kennicutt 1992b ) .
this limit is not unexpected for the bulge - dominated region of a galaxy .
we can also specify the details of the star formation process that correspond to the spectral synthesis models which were discussed in the previous section and illustrated in figures 4 and 5 . using the total magnitudes given in table 2 we find that : ( a ) for the best - fitting 0.72 gyr old burst model there is ( of course ) no current star formation and the total stellar mass involved is @xmath103 m@xmath44 , ( b ) for the 0.36 gyr old burst model the total mass is @xmath104 m@xmath44 , ( c ) for the 3.5 gyr old burst model the total mass is @xmath105 m@xmath44 , and ( d ) for the 3.5 gyr old @xmath84 model the total mass is @xmath106 m@xmath44 , where the initial star formation rate is 7.8 m@xmath44 yr@xmath9 and the star formation rate at @xmath15 is 0.24 m@xmath44 yr@xmath9 .
this is consistent with our upper limit on the star formation of 0.3 m@xmath44 yr@xmath9 from h@xmath1 .
we note that using total magnitudes ( table 1 ) results in stellar masses and star formation rates that are about a factor of two larger than when aperture magnitudes are used . _
3.4.1.4 . the radial brightness profile _
the radial brightness profile of g1 is also of interest . using the wiyn r - band data with the best seeing ( @xmath19 arcsec )
, we analyzed the radial brightness profile of g1 .
the observed light profile is the convolution of the true , two - dimensional light profile and the image point - spread function .
if the true profile is symmetric and the psf is well approximated by a gaussian , the convolution reduces to a simple one - dimensional integral ( see binney & tremain 1987 ) that makes it possible to compare a model radial brightness profile to the data in a straightforward manner .
the observed light profile of g1 is very close to symmetric , so we used this method .
elliptical isophotes were fitted to the brightness profile of g1 . the resulting radial brightness profile along the semi - major axis
is shown in figure 6 .
three models were fitted to the data : ( a ) a pure r@xmath107 ( bulge ) profile , ( b ) a pure exponential ( disk ) profile , and ( c ) a combination bulge - disk profile . as shown in the corresponding panels ( a - c ) of figure 6 , a pure - disk model is a poor fit , while a pure - bulge model gives better results .
the bulge - disk combination is clearly the best fit and is the preferred model , but it includes an extra free parameter .
this simple model and the small angular extent of g1 compared to the seeing make it difficult to pin down the bulge - to - total light ratio , the bulge half - light radius or the disk scale - length with accuracy , but our fits suggest that a moderate amount of each is necessary to fit the light profile well . owing to the need for some disk light , this result therefore seems reasonably consistent with the interpretation that g1 is an early - type `` dwarf spiral , '' which has been discussed as a recently - recognized new class of galaxy ( schombert et al .
1995 ) . in many respects
these galaxies resemble the h1-rich dwarf es , e / s0s , and s0s mentioned in 3.4.1.1 . according to schombert et al .
, the dwarf spirals are found only in the field and have total luminosities @xmath108 , diameters @xmath109 kpc , low central surface brightnesses @xmath110 b mag arcsec@xmath31 , and low h1 masses @xmath111 m@xmath44 ( converting the schombert et al .
values to our assumed cosmology with @xmath8 km s@xmath9 mpc@xmath9 ) .
this represents a class of objects which are currently rich in neutral gas , small in angular size , and low in surface brightness .
the central surface brightness of g1 ( @xmath112 b mag arcsec@xmath31 ) is somewhat brighter than the parameters given by schombert et al . _
3.4.1.5 . the h1 mass estimate _
g1 is displaced 5.7 arcsec from the quasar sight - line , corresponding to an impact parameter of @xmath17 kpc .
this is a relatively large impact parameter in relation to the apparent optical size of g1 ( figure 1 ) .
it is not unusual to find evidence for a relatively large impact parameter when trying to identify galaxies giving rise to quasar absorption - line systems , but it is somewhat surprising for a system and so it leaves open the possibility that the site of the absorption is an h1 cloud in a galaxy that is a companion to g1 .
however , consideration of this is beyond the scope of the available data , so we have only concentrated on the description of g1 here .
we note that if the @xmath26 absorber is assumed to have a uniform h1 column density of @xmath113 atoms @xmath3 and extend over a radius equal to the impact parameter , then its deduced h1 mass is @xmath114 @xmath115 m@xmath44 .
this h1 mass estimate is about an order of magnitude greater than quoted above for h1-rich dwarf es , e / s0s , s0s , or spirals .
the arm and jet are the two brightest patches of low surface brightness extended light near the quasar ( i.e. the `` fuzz '' ) . since we do not have a useful spectrum of any part of the fuzz , or data that can be used to derive a photometric redshift for these features , their redshifts can not be known with certainty . in principal
, the light could be from a combination of : ( 1 ) unrelated foreground or background objects at neither of the two damped redshifts , ( 2 ) the quasar host galaxy at @xmath116 , ( 3 ) the `` true '' galaxy at @xmath15 ( assuming that g1 at redshift @xmath15 does not have a sufficient h1 extent ) , and ( 4 ) the galaxy at @xmath6 . examination of the infrared images do suggest a combination of objects or perhaps a single disturbed object ( see figure 1 ) .
therefore , we will discuss each of these four possibilities .
\(1 ) contamination from unrelated objects is possible .
however , examination of the infrared images suggests that @xmath117% of the frame contains detectable infrared sources at the sensitivity of the k - band images .
therefore , if an unrelated object or objects are present near the quasar sight - line , this is an unlucky configuration , so we will not consider this possibility further .
\(2 ) there is a reasonable chance that some of the quasar s host galaxy light at @xmath116 is present in the infrared light surrounding the quasar .
the quasar , oi@xmath0363 , is optically luminous ( [email protected] , m@xmath118 and radio loud .
infrared studies of optically luminous , radio loud quasars in the h - band ( mcleod & reike 1994 ) show that it is not unusual for these types of quasars to have luminous host galaxies , e.g. , @xmath119 , and hst - wfpc2 studies at @xmath120 show that the host galaxies of radio loud quasars are almost exclusively ellipticals ( hamilton , casertano , & turnshek 2000 ) .
examination of the infrared oi@xmath0363 images after subtraction of the unresolved nuclear quasar light ( figure 1 ) indicates that the remaining light is not exclusively from an elliptical host centered on the quasar .
a reasonable deconvolution of the image suggests that the nuclear quasar component has @xmath121 ( @xmath122 ) while its host has an @xmath123 profile and @xmath124 ( @xmath125 ) , which is completely consistent with previous findings on radio loud quasar hosts .
nevertheless , after subtraction of the nuclear and host components some light remains between the arm and the jet . in any case
, the mere presence of the arm and the jet indicates that the surrounding light is not exclusively associated with an elliptical host .
an illustration of the light that we believe is dominated by the quasar elliptical host is contained within the ellipse drawn on figure 1a .
however , there are uncertainties with this deconvolution and better spatial resolution would be needed to accurately separate out the various components .
\(3 ) the possibility that some of the extended , low surface brightness light surrounding the quasar is due to the galaxy at @xmath15 is dismissed since g1 is identified as the galaxy . however , recall that the impact parameter of g1 is 5.7 arcsec or @xmath17 kpc , which may be large for a dwarf galaxy giving rise to absorption ( 3.4.1.5 ) .
\(4 ) from the arguments presented above ( 1 - 3 ) , we conclude that _ some _ of the light surrounding the quasar is from the galaxy at @xmath6 . owing to the approximately circular nature of the resolved light , regardless of whether a contribution from an elliptical host is removed , and the possibility of an actual spiral arm
, it may be that this galaxy is a nearly face - on spiral .
if this turns out to be the case , the low surface brightness dwarf spiral interpretation might in fact be more descriptive of the @xmath6 galaxy than for g1 at @xmath15 ( 3.4.1.4 ) . at the same time
, the considerable morphological structure surrounding the quasar leaves open the possibility that the @xmath6 galaxy is an irregular or an interacting system . in any case , if we want to place a conservative _ upper limit _ on the light from the @xmath6 galaxy it is appropriate to assume that _ all _ of the light surrounding the unresolved part of the quasar image ( including what we believe to be the quasar host ) belongs to it . with this assumption
we obtain an upper limit on the k - band luminosity of the galaxy at @xmath6 of @xmath126 . assuming the arm and jet are both at @xmath6 , the arm has @xmath127 with color r@[email protected] ( i.e. it is red and not detected in our r - band wiyn data ) and the jet has @xmath129 with a blue color of r@[email protected] .
this would imply that the jet is a site of relatively recent star formation , while the arm is an older population .
we estimate the diameter of the measurable light distribution to be @xmath130 arcsec , or @xmath131 kpc , and the separation from the quasar sight - line to be @xmath14 arcsec , or impact parameter @xmath13 kpc . as noted above , light contributed by an elliptical quasar host galaxy ( i.e. some light inside the ellipse shown on figure 1a ) might have to be subtracted from the measured upper limit on the galaxy s light in order to determine its true luminosity .
in fact , a reasonable deconvolution suggests that the quasar host contributes @xmath132% of the total measured light , in which case the luminosity of the @xmath6 galaxy is @xmath133 .
regardless of the amount of contribution from the host galaxy , the resulting interpretation is that the @xmath6 galaxy is a low surface brightness galaxy with a diameter of @xmath134 kpc at low impact parameter from the quasar sight - line .
the details can not be sorted out definitively without better data .
thus , the morphology of this low surface brightness dwarf galaxy remains uncertain @xmath66 ( another ? ) dwarf ( possibly face - on ) spiral , or an irregular , or an interacting system .
given the above indication of the putative @xmath6 galaxy s diameter , it is useful to estimate its h1 gas mass using some reasonable assumptions . if we simply assume that the circular h1 gas radius is the same as the extent of the galaxy seen in the k - band image ( i.e. radius r @xmath51 7 kpc ) and that the h1 column density is uniformly @xmath135 atoms @xmath3 over that extent , the deduced h1 mass is @xmath136 m@xmath44 .
this is a factor of 2 larger than the upper limit of lane et al .
( 2000b ) but , as discussed in 2.1 , some of the h1 21 cm emission searched for by lane et al .
( 2000b ) may have been missed since velocity intervals which included absorption had to be excluded in their analysis . at the same time , the assumptions used above may over - estimate the h1 mass since the impact parameter is evidently small and the h1 gas distribution may be patchy or fall off exponentially with impact parameter .
nevertheless , the order of magnitude of this estimate is of interest ; the total h1 mass is smaller than our estimate for the @xmath26 system ( 3.4.1.5 ) .
although there are still open questions which better data could address , the findings discussed here represent some of the first results on the _ detailed properties _ of galaxies .
perhaps fittingly , these results were determined from the two lowest redshift ( @xmath6 and @xmath137 ) confirmed classical systems known .
the results of our analysis indicate the following . the @xmath15 galaxy is a faint ( b@xmath138 , k@xmath139 ) neutral - colored ( b@xmath66k@xmath94 ) dwarf galaxy with k- and b - band luminosities of @xmath140 and @xmath141 , respectively .
its spectrum is that of an early type galaxy ( s0 or even e ) .
analysis of the 7-color ( ubrijhk ) data suggests that g1 s spectral energy distribution is well - fitted by an instantaneous single - burst model with the burst taking place 0.36 ( a@xmath142 ) to 3.5 ( a@xmath143 ) gyrs ago , or a 3.5 gyr old model with an exponentially decreasing star formation rate ( @xmath84 gyr ) and a@xmath144 .
its current rate of star formation is low , @xmath145 m@xmath44 yr@xmath9 . examination of the evolutionary history of these models leads us to conclude that the progenitor population of g1 , if seen near the population s formation epoch at @xmath18 , would have properties similar to the population of `` faint blue galaxies '' presently observed at moderate redshift .
analysis of the extended radial light profile from the wiyn r - band image of g1 indicates that its outer isophotes are better fit by an exponential than an r@xmath107 law , which offers some evidence for an early - type dwarf s0 or spiral interpretation .
the putative @xmath6 galaxy is likely to be all or part of the resolved light surrounding the quasar .
it is also a dwarf galaxy , and is inferred to have @xmath146 , with the higher value holding for the case where the quasar host galaxy at @xmath116 does not contribute to the detected resolved light .
there is considerable morphological structure in the light surrounding the quasar , including what we have called an arm and a jet .
the arm is red in color and if it is an _ actual _ arm , this suggests that the galaxy might be a low surface brightness dwarf spiral , but the presence of the other structure ( i.e. a blue colored jet with apparent structure ) leaves open the possibility that the galaxy is an irregular or an interacting system . in any case , the @xmath15 galaxy is more compact , while the inferred galaxy at @xmath6 has low surface brightness . for these galaxies the h1 gas masses deduced from the observed impact parameters and h1 column densities lie in the range @xmath147 m@xmath44 .
these estimates are crude because we do not know if the observed h1 is distributed like a face - on circular slab around the center of the identified galaxy ( like our calculation of the h1 mass assumed ) or whether the h1 gas is exponentially distributed or clumped in giant clouds associated with the identified galaxy . in any case
, it would clearly have been much more difficult to deduce these properties from similar observations if these same objects were at moderate to high redshift . however , a collection of such detailed results , used in conjunction with statistical results on the incidence , cosmological h1 gas mass density and column density distribution of absorption systems ( rao & turnshek 2000 ) , will eventually be needed to accurately reveal the evolution of the bulk of the neutral gas in the universe ( as tracked by quasar absorption systems ) and the corresponding evolutionary relationship to galaxies ( i.e. the associated stellar light ) .
one qualitative question presently of interest is to determine if the population is slowly evolving from @xmath148 to @xmath149 , as has been recently suggested by pettini et al .
( 1999 ) and rao & turnshek ( 2000 ) on the basis of their metallicities and statistics , respectively .
it would seem that there has to be some significant evolution in the population from high to low redshift if the results of rao & briggs ( 1993 ) at @xmath150 are correct .
the rao & briggs ( 1993 ) study places almost all of the local h1 gas in luminous spirals .
present evidence suggests that luminous galaxies were approximately in place by @xmath151 , with less luminous galaxies showing more evolution at lower redshift ( e.g. i m et al . 1996 , 1999 ) .
therefore , one could envisage a scenario where luminous galaxies contribute little to the detections , except locally , while an evolving population of dwarfs and low surface brightness galaxies cause the bulk of the detections at @xmath152 .
indeed , the present searches for low - redshift dla galaxies ( le brun et al 1997 ; turnshek et al .
2000 ) suggest that a significant fraction of them are dwarf and low surface brightness galaxies , and the possible connection to the evolving moderate redshift `` faint blue galaxy '' population mentioned earlier is intriguing .
clearly the statistical survey results and the low - redshift follow - up studies of dla systems need to be extended and improved to address these evolutionary issues properly .
it is also of interest to consider selection effects more carefully as well as design experiments which will reach to lower h1 masses in searches for large quantities of h1 gas not associated with luminous spirals .
the current 21 cm emission surveys are sensitive to @xmath153 m@xmath44 locally .
regardless of the outcome of such new programs , the results reported here are consistent with our earlier findings ( rt98 ) and serve to emphasize that absorbing galaxies are not exclusively luminous spirals .
they are drawn from a wide variety of types of gas - rich galaxies .
we wish to thank dr .
frank briggs for collaboration on parts of this program and drs .
bolzonella , pello , and miralles for making their spectral synthesis code _ hyperz _ publicly available .
we wish to also thank the support astronomers at the irtf , mayall , mdm , and wiyn telescopes for their assistance .
this work was supported in part by an nsf grant and a nasa - ltsa grant .
turnshek , d. a. , rao , s. , lane , w. , monier , e. , nestor , d. , bergeron , j. , & smette , a. 2000 , to appear in _ gas & galaxy evolution _
, asp conference series , eds .
j. hibbard , m. rupen , and j. van gorkom , in press ( astro - ph/0009096 ) cclc 4.0 in u & mdm & 1999 feb 20 & 60 + `` & '' & 1999 nov 13 & 90 + `` & '' & 1999 nov 14 & 60 + b & wiyn & 1997 nov 5 & 30 + `` & '' & 1998
nov 25 & 15 + r & `` & 1997 nov 5 & 45 + '' & `` & 1999 jan 18 & 15 + i & '' & 1997 dec 5 & 50 + j & irtf & 1998 dec 13 & 30 + `` & '' & 1999 apr 29 & 63 + h & `` & 1998 dec 13 & 30 + '' & `` & 2000 mar 8 & 63 + k & '' & 1998 dec 12 & 130 + `` & '' & 1998 dec 14 & 132 + `` & '' & 1999 apr 28 & 63 + `` & '' & 2000 mar 6 & 126 ccccccc 6.0 in u & 24.7 & 23.67 ( 0.15 ) & 22.8 ( 0.15 ) & 23.8 ( 0.3 ) & & + b & 25.7 & 23.46 ( 0.05 ) & 22.7 ( 0.05 ) & 23.4 ( 0.2 ) & & + r & 27.0 & 21.49 ( 0.07 ) & 20.8 ( 0.05 ) & 21.5 ( 0.1 ) & 25.1 ( 0.2 ) & + i & 24.1 & 21.07 ( 0.12 ) & 20.4 ( 0.1 ) & 21.0 ( 0.1 ) & & + j & 22.8 & 19.82 ( 0.12 ) & 18.9 ( 0.1 ) & 19.9 ( 0.1 ) & 22.1 ( 0.5 ) & + h & 21.0 & 19.38 ( 0.12 ) & 18.7 ( 0.1 ) & 19.6 ( 0.1 ) & & + k & 22.5 & 18.52 ( 0.12 ) & 17.8 ( 0.1 ) & 18.3 ( 0.1 ) & 21.9 ( 0.2 ) & 21.5 ( 0.2 ) | new optical and infrared observations along the sight - line toward the quasar oi@xmath0363 ( 0738 + 313 ) are presented and discussed . excluding systems which lack confirming uv spectroscopic observations of the actual ly@xmath1 line
, this sight - line presently contains the two lowest - redshift classical damped ly@xmath1 ( ) quasar absorption line systems known ( i.e. with @xmath2 atoms @xmath3 ) , one at @xmath4 and the other at @xmath5 .
our new observations suggest identifications for the galaxy counterparts of these absorption - line systems .
the @xmath6 galaxy appears to be an extended low surface brightness galaxy which is easily visible only in infrared images and shows rich morphological structure .
assuming there is no contribution from the quasar host galaxy , we place an upper limit on the k - band luminosity of the @xmath6 galaxy of @xmath7 ( assuming a cosmology with @xmath8 km s@xmath9 mpc@xmath9 , @xmath10 , and @xmath11 ) .
more realistically , a subtraction of the quasar nuclear and host light yields @xmath12 .
the impact parameter between the galaxy and quasar sight - line is very small , @xmath13 kpc ( @xmath14 arcsec ) , which makes measurements difficult . the @xmath15 galaxy is an early - type dwarf with a k - band luminosity of l@xmath16 at impact parameter @xmath17 kpc .
its colors are neutral and consistent with star formation models suggesting its formation epoch was less than a few gyr ago ( i.e. @xmath18 ) .
thus , it is conceivable that its progenitor originated from the population of `` faint blue galaxies '' seen at moderate redshifts . in general , these results serve to support mounting evidence that galaxies are drawn from a wide variety of gas - rich galaxy types . |
supersymmetry ( susy ) is one of the most promising scenarios of physics beyond the standard model ( sm ) and predicts new particles observable at the present or next generation of colliders .
it solves the naturalness problem by taming the otherwise quadratically divergent quantum correction to the higgs boson mass , makes the three gauge forces unify , and opens up a way to ultimate unification of all the fundamental interactions including gravity @xcite .
if susy is the case , we will be able to probe physics at the unification scale by studying its breaking pattern through various measurements of susy particles some of which will have been discovered at the tevatron or the lhc .
one of primary goals for an international linear collider ( ilc ) is hence determination of susy or susy - breaking parameters through precision measurements of masses and couplings of those susy particles .
the pair production of lighter charginos ( @xmath7 ) is one of the most important processes in this respect since the @xmath3 is relatively light in most of the parameter space and hence expected to be studied in the early stage of the ilc project .
measurements of its mass and couplings at the ilc have not only been discussed but , for some typical sets of susy(-breaking ) parameters , also simulated in detail @xcite .
all of the cases studied in the past , however , assume that the lighter chargino would decay directly into a final state containing the lightest supersymmetric particle ( lsp ) , which is the lightest neutralino ( @xmath6 ) in most of the susy parameter space . at large @xmath1 values , however , @xmath8-@xmath9 mixing might generate a large mass splitting between @xmath10and the other sleptons ( @xmath11 , @xmath12 , @xmath13 , @xmath14 ) , yielding a particular mass hierarchy of @xmath15 . if @xmath16 is gaugino - like , this means that the @xmath3 will decay at 100% of the times in the cascade mode : @xmath17 followed by @xmath5 .
consequently , the experimental signature of the chargino pair production will be an acoplanar @xmath18 pair with significant missing energy carried away not only by the lsp s from the @xmath2 decays but also by the final - state neutrinos from the @xmath3 and @xmath18 decays .
this would make it significantly more difficult to determine the chargino mass .
this paper describes results from our simulation studies of the pair production of the lighter charginos and their subsequent cascade decays : @xmath19 in a particular scenario of the minimal supersymmetric standard model ( mssm ) , where @xmath20 gev/@xmath21 and @xmath22 gev/@xmath21 as in refs .
it should be noted that signals from the @xmath23-@xmath3 co - annihilation region are recently of great interest .
the region is characterized by a difference in the @xmath23 mass and the lightest neutralino mass of about 5 - 15 gev/@xmath21 .
this small mass difference allows the @xmath23 to co - annihilate in the early universe along with the neutralinos in order to produce the current amount of cold dark matter density of the universe measured by wmap @xcite .
however , the experimental signature is very challenging
. we will not address this signature , because it requires to detect electron / positron from two - photon process in the forward region .
thus it requires a different optimization of the event selection criteria @xcite .
in simulating the signal and various background events , we use the helas library @xcite to calculate the helicity amplitudes , the bases / spring package @xcite to generate the final - state partons , and tauola @xcite to decay the final - state @xmath18 leptons if any , so as to take into account possible spin correlations of intermediate heavy partons , effects of beamstrahlung and subsequent beam - energy spread @xcite , and finite @xmath18 polarizations .
the generated events are then processed through a fast simulator of a detector model @xcite , which includes a central tracker , electromagnetic and hadron calorimeters , and muon drift chambers .
this simulator smears charged - track parameters in the central tracker with parameter correlation properly taken into account , and simulates calorimeter signals as from individual segments .
although @xmath18 polarization varies in general with the susy parameters and would thus influence the property of the final states , we fix the @xmath18 polarization for the signal process at the nominal value of 0.026 for simplicity . for susy events ,
we choose our reference susy point to have @xmath24 , @xmath25 gev , m@xmath26 gev , and m@xmath27 = 180.0 gev , where @xmath28 , m@xmath29 , and m@xmath27 are the higgsino mixing mass , universal scalar mass at gut scale , and su(2)@xmath30 gaugino mass parameters , respectively .
we take m@xmath31 so that it satisfies gut relation m@xmath32m@xmath27 .
it corresponds to @xmath33 gev/@xmath21 ( @xmath34 ) , @xmath35 , and @xmath36 .
notice that @xmath37 is the next to the lightest susy particle ( nlsp ) and decays to @xmath38 at 100% of the time .
we assume in what follows that @xmath39 and @xmath40 will have been measured below the threshold of chargino pair production , say at @xmath41 = 310 gev , with uncertainties expected for @xmath42 @xcite . in order to evaluate to what extent we can determine the chargino mass in this setting
, we then vary the @xmath3 mass around 171.0 gev/@xmath21 , while keeping the @xmath2 and @xmath6 masses unchanged , by appropriately adjusting the values of susy(-breaking ) parameters gaugino mass parameters ( m@xmath43 and m@xmath44 ) that has been assumed for our reference point . ] and see how the observable distributions respond to the change . as for the ilc parameters , we assume @xmath45 gev , @xmath46 , and an electron beam polarization of @xmath47 ( left handed ) .
the @xmath48 production cross section at @xmath47 is 20 times larger than that at @xmath49 ( right handed ) .
furthermore , it can suppress the cross section for the @xmath50 process that yields almost the same final states as our signal process .
in our setting , the chargino pair production results in a final state consisting of a @xmath18-pair with large missing energy .
the five major backgrounds for the signal event are considered : ( i ) @xmath37-pair production : @xmath51 , ( ii ) neutralino pair production : @xmath52 , ( iii ) diboson production : @xmath53 , @xmath54 ( @xmath55 ) or @xmath56 , ( iv ) single boson production : @xmath57 and @xmath58 with @xmath59 or @xmath60 , and ( v ) two - photon @xmath18-pair production : @xmath61 . as for the neutralino pair production , the su(2)@xmath30 relation between @xmath62 and @xmath3 masses is assumed .
we also assume the @xmath62 mass is measured via other processes such as @xmath63 .
we ignore the uncertainty in the @xmath62 mass measurement , because the effect in @xmath3 mass measurement is found to be negligible .
mass determination accuracy is estimated to be @xmath64 gev/@xmath21 if the @xmath62 mass fluctuation is @xmath65 gev/@xmath21 . ]
notice that the cross section times branching fraction for @xmath66-pair production followed by @xmath67 decays is , for instance , eight times higher than that of the signal at our susy reference point .
we had thus better avoid events with purely leptonic @xmath18 decays .
notice also that the two @xmath18-jets decay will hereafter be called a @xmath18-jet even if it contains only one charged particle . ] in the susy events are likely to be more acoplanar than those from the sm backgrounds .
these observations lead us to the following selection criteria to improve signal - to - background ratio : there should be no @xmath68 candidates positively identified in the detector simulator ; jet clustering has to yield two jets with @xmath69 for @xmath70 @xcite ; @xmath71 , where @xmath72 and @xmath73 are the charge and the momentum of @xmath74-th jet with @xmath74 = 1(2 ) corresponding to the higher ( lower ) energy jet ; @xmath75gev/@xmath21 ; the missing transverse momentum @xmath76gev/@xmath77 ; @xmath78 or @xmath79 , where @xmath80 and @xmath81 is the visible momentum vector calculated with both tracker and calorimeter information ; thrust @xmath82 ; acoplanarity angle @xmath83 @xmath84 .
cut ( f ) effectively rejects the @xmath85 background , because the @xmath18 lepton from @xmath86 tends to go in the parent @xmath66-boson momentum direction .
we tabulate the resultant selection efficiencies after each cut for the signal and the background processes at the reference susy point in table [ tab : eventsel ] .
total selection efficiency ( @xmath87 ) for the @xmath88 signal at the reference point is @xmath89 .
we see that the selection criteria are effective at suppressing the sm backgrounds , yielding a signal - to - background ratio of 1.7 . lrrrrrrrr process & @xmath48 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath95 & + & ( ref . ) & ( ref . ) & ( ref . ) & & & & & @xmath96 + @xmath97 ( fb ) ( @xmath47 ) & 322 & 78 & 44 & 2304 & 83.6 & 1876 & 92.8 & 5.62@xmath98 + ( a ) @xmath68 removal & 0.403 & 0.424 & 0.398 & 0.0747 & 0.0623 & @xmath99 & 0.0437 & @xmath100 + ( b ) n@xmath101 = 2 & 0.351 & 0.308 & 0.348 & 0.0392 & 0.0522 & @xmath102 & 0.0377 & @xmath103 + ( c ) jet angle cut & 0.240 & 0.218 & 0.251 & 0.0122 & 0.0255 & @xmath104 & 0.0239 & @xmath105 + ( d ) @xmath106 gev/@xmath21 & 0.220 & 0.187 & 0.238 & 0.0118 & 0.0229 & @xmath107 & 0.0232 & @xmath105 + ( e ) @xmath108 gev/@xmath77 & 0.182 & 0.147 & 0.198 & 0.0085 & 0.0208 & @xmath109 & 0.0178 & @xmath110 + ( f ) @xmath111 or & & & & & & & + @xmath112 & 0.175 & 0.142 & 0.185 & 0.0054 & 0.0200 & @xmath113 & 0.0161 & @xmath114 + ( g ) thrust @xmath115 & 0.173 & 0.141 & 0.172 & 0.0043 & 0.0200 & @xmath116 & 0.0160 & @xmath117 + ( h ) acoplanarity @xmath118 & 0.158 & 0.137 & 0.158 & 0.0031 & 0.0186 & @xmath119 & 0.0136 & @xmath120 + @xmath87 & 0.158 & 0.137 & 0.158 & 0.0031 & 0.0186 & @xmath119 & 0.0136 & @xmath121 + ( generated events ) & 0.4 m & 0.4 m & 0.6 m & 1 m & 1 m & 1.4 m & 1 m & 8 m + # of expected events & & & & & & & + for @xmath122 = 200 fb@xmath123 & 10175 & 2137 & 1390 & 1428 & 311 & 34 & 252 & 315 + + as for the @xmath124 event where quark jets are mis - identified as @xmath18-jets , the total selection efficiency is small ( less than @xmath125 ) and the expected number of the @xmath124 events is 2.5 . we do not include this process in table [ tab : eventsel ] . we also show the signal selection efficiencies for different chargino masses in table [ tab : eventsel1 ] .
the @xmath2 and @xmath6 masses are fixed to be 152.7 gev/@xmath21 and 86.4 gev/@xmath21 , respectively .
lrrrrr @xmath126 ( gev/@xmath21 ) & 161.0 & 166.0 & 171.0 & 176.0 & 181.0 + & & & ( ref . )
& & + @xmath127 ( gev/@xmath21 ) & 8.3 & 13.3 & 18.3 & 23.3 & 28.3 + @xmath97 ( fb ) & 433 & 378 & 322 & 268 & 214 + @xmath87 & 0.156 & 0.157 & 0.158 & 0.158 & 0.158 + + the statistical uncertainty in the efficiency for each point is better than 0.1% .
the efficiency depends on the @xmath3 mass only very weakly as long as @xmath128 is large enough to deposit a jet energy of @xmath129 5 gev in the detector .
if @xmath6 and @xmath2 masses are predetermined , the @xmath18-jet energy distribution can be used to determine the @xmath3 mass , since the shape of the distribution , especially around the endpoint , depends on @xmath130 . as @xmath130 increases , the peak of the distribution shifts to the higher energy side , while the endpoint moves to the lower energy side .
this can be seen in figure [ fig : final2 ] , which plots the @xmath18-jet energy distributions for various @xmath130 values . in order to determine the chargino mass using the @xmath18-jet energy distribution , we first prepare a parametrized @xmath18-jet energy distribution ( a template ) as a function of @xmath131 in the following manner : ( i ) the @xmath18-jet energy distribution , including the backgrounds , is fit to a polynomial function by minuit @xcite for each of high statistics samples with different @xmath3 masses ; ( ii ) the fitting parameters are then parametrized as a function of @xmath131 . although the @xmath18-jet energy distribution may depend on the remaining free parameters such as @xmath14 mass , @xmath28 parameter , and @xmath18 polarization , those are fixed in the present study .
we then compare the @xmath18-jet energy distribution for a monte carlo sample corresponding to 200 fb@xmath123 with the template and calculate @xmath132 .
figure [ fig : delw1ss ] shows @xmath133 for the reference point ( @xmath34 = 171.0 gev/@xmath21 ) .
the fitted value of the @xmath3 mass is @xmath134 gev/@xmath21 .
this is about 1@xmath97 off the reference mass of 171.0 gev/@xmath21 .
we prepare approximately 100 statistically independent samples of @xmath135 and @xmath136 events for the reference point .
each sample is also corresponding to 200 fb@xmath123 .
the @xmath132fitting procedure provides consistently a @xmath137 uncertainty of @xmath138 gev/@xmath21 for each sample .
however the fluctuation of @xmath3 mass at the @xmath139 is @xmath140 gev/@xmath21 around the @xmath34 .
therefore , we assign the total uncertainty in the @xmath3 mass measurement to be @xmath141 gev/@xmath21 ( quadratic sum of two sources ) . the above estimate ignores the uncertainties in the input @xmath2 and @xmath6 masses .
these uncertainties will affect the accuracy in the @xmath3 mass measurement , because the differences between @xmath2 , @xmath6 , and @xmath3 masses determine the shape of @xmath18-jet energy distribution .
the higher endpoint , in particular , depends on the difference ( @xmath130 ) between @xmath2 and @xmath3 masses . in order to evaluate the impact of the uncertainties on the @xmath3 mass determination , we repeat the study for @xmath142 , varying @xmath143 and @xmath144 .
figure [ fig : contchi ] shows a contour plot of @xmath145 in the @xmath143 - @xmath144 plane .
the center point of each @xmath146 contour is at @xmath147 and @xmath148 .
the contours clearly demonstrate a strong positive correlation between @xmath144 and @xmath143 , implying that the shape of the @xmath18-jet energy distribution is mainly controlled by their difference , as long as the @xmath3 mass is fixed .
it is worth noting that the smallest ellipse , which is the expected 1@xmath97 bound from the @xmath2-pair production study with 100 fb@xmath123 cited from ref .
@xcite , has a quite similar correlation and is well contained in the contour corresponding to @xmath149 .
the aforementioned expected accuracy of @xmath141 gev/@xmath150 for the @xmath3 mass will hence essentially be unaffected by the inclusion of the uncertainties in the @xmath6 and @xmath2 masses from the lower energy @xmath2-pair study .
the production cross section times branching ratio is experimentally estimated by @xmath151 , where @xmath152 is the number of observed events , @xmath153 the number of expected background events , @xmath154 the integrated luminosity , and @xmath87 the selection efficiency .
it should be noted that @xmath87 weakly depends on @xmath131 as shown in table [ tab : eventsel1 ] : the efficiency varies by only 0.6% over the mass range of @xmath155 gev/@xmath21 from @xmath34 .
this implies the possibility of extracting the cross section times branching fraction ( @xmath156 fb ) independently of the input @xmath3 mass , provided that @xmath157 fb@xmath123 ( the luminosity is assumed to be measured with an uncertainty of 1% @xcite ) , @xmath158 events ( the background contribution is fully estimated ) , and @xmath159 ( the uncertainty includes the weak dependence of @xmath87 on @xmath131 ) .
this indicates that we can determine the production cross section times branching fraction to an accuracy of 2% even without any precise @xmath131 study .
at large @xmath1 values , the @xmath2 mass might be lighter than the @xmath3 mass within the mssm framework . in this case
, the @xmath17 decay will be dominant and that will be followed by @xmath160 .
we have thus studied the final state consisting of @xmath161 + large missing energy as a possible signature of the chargino pair production at a future @xmath162 linear collider .
we proposed to use the @xmath18-jet energy distribution to measure the @xmath3 mass in the large @xmath1 scenario .
since this scenario leads @xmath6 and @xmath2 masses to be lighter than @xmath3 mass and opens the @xmath2 pair production mode with lower energy of @xmath3 pair production , the @xmath6 and @xmath2 masses are likely to be determined in the lower @xmath3 production threshold energy operation of the collider .
the measurements of 171 gev/@xmath150 chargino could then be made with an accuracy better than 1 gev/@xmath150 , provided that the @xmath6 and @xmath2 masses will have been determined through a @xmath2-pair production study with 100 fb@xmath123 .
since the event selection efficiency does not significantly depend on the @xmath3 mass , the uncertainty in the production cross section shall be dominated by uncertainties in the selection efficiency and the luminosity measurement .
we expect to determine the cross section to an accuracy of 2% at the reference point despite the cascade decay mode .
the authors would like to thank r. arnowitt for carefully reading the manuscript .
is supported by the grant for kinki university , while k.f .
is supported in part by japan - europe ( uk ) research cooperation program and jsps - cas scientific cooperation program under the core university system , t.k . and v.k .
department of energy grant de - fg02 - 95er40917 , and m.m.n . by the grant - in - aid for science research , ministry of education , science and culture , japan ( no.14540260 and no.14046210 ) , respectively .
t. tsukamoto , k. fujii , h. murayama , m. yamaguchi , and y. okada , ; j. feng , m. peskin , h. murayama , and x. tata , ; s.y .
choi , m. guchait , j. kalinowski , and p.m. zerwas , ; j.a .
aguilar - saavedra , et al . , _ tesla technical design report , part iii : physics at an @xmath162 linear collider _ , desy 2001 - 11(2001 ) ; t. abe et al.(american linear collider working group ) , _ linear collider physics resource book for snowmass 2001 , part2 : higgs and supersymmetry studies _ , slac - r-570(2001 ) . | the lighter chargino pair production ( @xmath0 ) is one of the key processes for determination of supersymmetric parameters at a linear collider .
if @xmath1 is a large value , the lighter stau ( @xmath2 ) might be lighter than the @xmath3 , while the other sleptons stay heavier .
this case leads to a cascade decay , @xmath4 followed by @xmath5 , where @xmath6 is the lightest supersymmetric particle .
this paper addresses to what extent this cascade decay may affect the measurements of the chargino mass and its production cross section . , , , , supersymmetry , linear collider |
in the last two decades a great attention has been devoted to the understanding of the microscopic underpinnings of the various formulations of the second law of thermodynamics on the basis on exact non - equilibrium identities , collectively known as fluctuation relations @xcite .
for example , when applied to a cyclic process occurring in a system in contact and initially in equilibrium with a bath of temperature @xmath5 , the jarzynski identity @xmath6 , an exact theorem in hamiltonian ( classical or quantum ) mechanics @xcite , implies that the average work done on the system is non - negative @xmath7 , namely no energy can be extracted from the bath ( kelvin formulation of the second law ) denotes the difference between the free energy of the ( not necessarily reached ) state of thermal equilibrium corresponding to the end - point of the protocol and the initial equilibrium state . for a cyclic protocol @xmath8 .
@xmath9 denotes boltzmann s constant . ] .
likewise , clausius formulation , @xmath10 follows from a similar identity @xmath11 , that applies when a driven system is brough into contact with various thermal baths @xcite .
when studying the heat @xmath12 flowing from a hot bath to a cold bath , with no external driving , the exchange fluctuation relation @xcite implies the identity @xmath13 which in turn implies @xmath14 , saying that heat , on average , can not flow from cold to hot : yet another equivalent formulation of the second law .
similarly , sinitsyn @xcite derived a nonequilibrium identity for heat engines , @xmath15 , leading to carnot s formulation : no heat engine can overperform the carnot engine , @xmath16 , see also ref . @xcite .
these derivations are based on the assumption that the engine dynamics is described by a stochastic jump process obeying detailed balance .
sinitsyn however suggested @xcite that a more general hamiltonian derivation should be possible as well .
here we follow that suggestion and provide a proof of the fluctuation relation for heat engines under the assumption that device and baths obey the laws of quantum mechanics .
we further extend that result to the case of reverse operation , namely for quantum refrigerators .
a number of proposals have been recently made of quantum engines and refrigerators , see , e.g. @xcite .
the fluctuation relations presented here constitute a basic theoretical tool for the study of those and similar devices .
, with two thermal reservoirs of temperature @xmath17 , respectively , and outputs the energy @xmath2 .
the arrow directions reflect the sign convention adopted in this work .
heat is positive when flows out of a reservoir , work is positive when done on the external body .
when the device operates as a heat engine it is @xmath18 .
when the device operates as a refrigerator it is @xmath19 . ]
consider a heat engine , fig .
[ fig : fig1 ] .
a device undergoes a cyclic process during which it exchanges the heats @xmath20 with two thermal reservoires at temperatures @xmath0 , respectively , and outputs energy in the form of work @xmath2 .
we treat the reservoires and device as a compound quantum mechanical system .
its hamilton operator reads @xcite : @xmath21 here @xmath22 is the hamiltonian of the device .
it depends on time through the parameter @xmath23 accounting for the coupling with the external work source / sink . the device can be for example a gas in a cylindrical container of volume @xmath23 .
the device undergoes a cycle of duration @xmath24 , i.e. @xmath25 . the two thermal reservoires have hamiltonians @xmath26 , while @xmath27 denote their coupling to the device .
the time - dependent real coefficients @xmath28 denote the degree to which each coupling is switched on .
following @xcite we consider the case when @xmath29 and @xmath30 with @xmath31 the heaviside step function .
accordingly the device is prepared at @xmath32 in a state of thermal equilibrium with the cold bath . assuming the coupling @xmath33 is weak , the initial state is given , up to second order corrections in the coupling , by @xcite : @xmath34 where @xmath35
, @xmath9 is boltzmann constant , @xmath36 , with @xmath37 denoting the trace over the device , hot bath , and cold bath respective hilbert spaces . according to the two - measurement scheme @xcite , at time
@xmath38 projective measurements are performed of the energy of the device @xmath39 , the energy of the cold bath @xmath40 , and the energy of the hot bath @xmath41 .
let @xmath42 denote the respective outcomes .
note that @xmath43 and @xmath41 commute with each other , hence they can be simultaneously measured .
the device+baths then evolves up to time @xmath24 according to the unitary evolution generated by the total hamiltonian ( [ eq : htot ] ) . at the final time ,
@xmath44 , joint projective energy measurements of @xmath45 and @xmath41 are again performed with outcomes @xmath46 . under the assumption that the total hamiltoinan @xmath47 is time - reversal invariant at each time @xmath48 , the joint probability @xmath49 of observing the energy changes @xmath50 , @xmath51
, @xmath52 obeys the fluctuation relation @xcite : @xmath53 where @xmath54 is the probability of measuring @xmath55 in the reverse operation of the machine @xmath56 starting in the initial state @xmath57 ( recall that we consider a cyclic driving , @xmath58 ) .
the proof of eq .
( [ eq : ft ] ) crucially depends on the unitarity and time reversal symmetry of the system+baths compound .
because of energy conservation the work output @xmath2 equals the negative total energy change in the device+baths , hence , neglecting the weak coupling energies : @xmath59 thus we arrive at the fluctuation relation for quantum heat engines : @xmath60 where @xmath61 is the joint probability of increasing the system energy by @xmath62 , exchanging heat @xmath3 and performing the work @xmath2 , and @xmath63 is the joint probability of increasing the system energy by @xmath64 , exchanging heat @xmath65 and performing the work @xmath66 in the reverse process . note that the r.h.s . of eq .
( [ eq : frqhe ] ) does not depend on @xmath62 .
this cancellation was already pointed out in @xcite , and crucially depends on the fact that the device is prepared in equilibrium with the cold bath .
multiplying eq .
( [ eq : frqhe ] ) by @xmath67 and integrating out @xmath68 we obtain the nonequilibrium identity : @xmath69 invoking now jensen s inequality we obtain : @xmath70 note that @xmath71 .
assuming that the machine operates as a heat engine , it is @xmath72 , hence we obtain @xmath73
it is interesting to see what happens when the machine works as a refrigerator .
we now consider the case @xmath74 , and prepare the device in a state of thermal equilibrium with the hot bath at time @xmath32 : @xmath75 repeating the same steps as above one arrives at : @xmath76 hence : @xmath77 assuming the engine is working as a refrigerator , it is @xmath78 , hence we arrive at : @xmath79
invoking the classical version of the fluctuation relation @xcite , which reads exactly as the quantum one , the same results can be obtained within the classical hamiltonian formalism as well .
note that the interpretation of @xmath3 @xmath80 as heat is only possible under the assumption that the couplings @xmath81 are weak , at least at the times @xmath82 of the two measurements @xcite . as already pointed out by sinitsyn @xcite , eq .
( [ eq : frqhe ] ) includes the work fluctuation relation for cyclic processes , and the exchange fluctuation theorem as special cases .
the work fluctuation theorem for cyclic processes emerges when @xmath83 , namely the system is in contact with a single bath . in this case one
obtains @xmath84 .
this same relation is also obtained , with @xmath85 , if the coupling to the hot bath is switched off , @xmath86 , or if both couplings are switched off @xmath87 .
note that the quantum heat engine fluctuation relation , eq .
( [ eq : integratedfrqhe ] ) , remains valid also if only the coupling to the cold bath is switched off .
this evidences that what counts the most in obtaining it is the initial preparation of the system in equilibrium with the _ cold _ bath , and the ability to interact with a hotter bath ( vice - versa for a refrigerator ) .
the exchange fluctuation relation @xmath88 emerges when there is no device , hence no driving , @xmath89 , and no @xmath62 , namely @xmath90 .
it is important to notice that in a real implementation , where the driving @xmath91 is a periodic function of time , the system typically enters into a non - equilibrium steady state , in which at the beginning of each cycle the system is not necessarily in equilibrium with the cold ( or hot ) bath .
this poses no problem regarding the validity of the here presented results , because all that is necessary for them to hold is that the system is in equilibrium with the cold ( or hot ) bath at the beginning of the driving , i.e. at the beginning of the very first cycle only .
the physical meaning of the bound in eq .
( [ eq:2ndlaw ] ) is that when repeating the cycle a very large number @xmath92 of times , the total work output @xmath93 over the total heat input @xmath94 can not overcome carnot efficiency .
note that the efficiency @xmath95 of each single operation is a stochastic quantity obeying the statistics @xmath96 where @xmath97 denotes dirac s delta function .
in general neither the variable @xmath98 nor its average @xmath99 ( which can well differ from @xmath100 due to possible correlations between @xmath3 and @xmath2 ) are bounded by carnot s efficiency . for recent investigations on the statistics of efficiency
see , e.g. @xcite .
fruitful discussions with prof .
r. fazio are gratefully acknowledged .
this research was supported by a marie curie intra european fellowship within the 7th european community framework programme through the project nequflux grant n. 623085 and by the cost action mp1209 `` thermodynamics in the quantum regime '' .
10 url # 1#1urlprefix[2][]#2 seifert u 2008 _ eur . phys .
j. b _ * 64 * 423431 | at the very foundation of the second law of thermodynamics lies the fact that no heat engine operating between two reservoires of temperatures @xmath0 can overperform the ideal carnot engine : @xmath1 .
this inequality follows from an exact fluctuation relation involving the nonequilibrium work @xmath2 and heat exchanged with the hot bath @xmath3 . in a previous work [ sinitsyn n a , j. phys .
a : math . theor .
* 44 * ( 2011 ) 405001 ] this fluctuation relation was obtained under the assumption that the heat engine undergoes a stochastic jump process . here
we provide the general quantum derivation , and also extend it to the case of refrigerators , in which case carnot s statement reads : @xmath4 . |
since the pioneering work by anderson and fazekas @xcite the spin-1/2 antiferromagnetic heisenberg model on the triangular lattice has been considered one of the most promising candidates for a spin liquid phase in a frustrated antiferromagnet . however several numerical studies @xcite have all consistently confirmed that in the isotropic triangular lattice the classical magnetically ordered state appears stable .
nevertheless the ordered moment is found considerably smaller than the classical value , @xcite suggesting that the model is very close to a quantum critical point , @xcite namely to a phase where the long range antiferromagnetic order is completely destroyed .
this picture is supported by the recently established result that in the quantum dimer model on the triangular lattice geometry @xcite a spin liquid phase is stable , a result that is particularly important because for instance the same model on the square lattice displays only non spin liquid phases with broken translation symmetry .
@xcite recent experiments on two different materials @xcite have renewed the interest in the spin-1/2 antiferromagnetic heisenberg model on the triangular lattice described by the following hamiltonian : @xcite @xmath9 where @xmath10 is a spin 1/2 located at site @xmath11 on the triangular lattice , @xmath12 ( @xmath13 ) indicates nearest neighbor sites along the chain ( between different chains ) , and the corresponding antiferromagnetic couplings are denoted by @xmath5 and @xmath6 ( see fig .
[ lattice ] ) .
clearly the anisotropy increases the quantum fluctuations in this model as for @xmath14 the hamiltonian @xmath15 reduces to a system of uncoupled one dimensional ( 1d ) chains , implying spin fractionalization and no antiferromagnetic order . in this limit
the spin one excitations should form a broad two - spinon continuum of states as predicted theoretically , @xcite and indeed several experiments by inelastic neutron scattering have revealed this non trivial spin dynamics .
@xcite and @xmath16 denote the primitive translational vectors .
a lattice constant is set to be one . ] in a series of experiments by coldea , _ et al .
_ , @xcite the low energy spin dynamics of @xmath0 have been studied systematically and found to be described essentially by the model hamiltonian ( [ model ] ) with anisotropic exchange interaction @xmath17 . in their experiments , @xcite there are two facts supporting the existence of a spin liquid phase in this system : i ) a spin spiral phase appears at temperatures below @xmath18 k , which is about the same order of magnitude as the inter - plane coupling @xmath19 in the third spatial direction , and therefore sizably smaller than the two dimensional ( 2d ) couplings @xmath5 ( @xmath20 mev ) and @xmath6 .
in fact , true long range order in 2d is possible only at zero temperature and a finite @xmath21 is due to the inter plane coupling @xmath22 which allows to cut - off the logarithmically divergent quantum fluctuations in 2d .
a finite @xmath21 is therefore expected to be of order @xmath23 , which appears slightly off from the observed @xmath24 k. ii ) at temperatures larger than @xmath21 ( or by applying an external magnetic field ) , the inelastic neutron scattering experiments @xcite found that the line shape of the spectrum consists of a broad continuum , which is in contrast to the expected behavior of a magnetically ordered state , but is instead similar to the broad two - spinon continuum expected in the 1d systems mentioned above . therefore these experiments suggest that spin fractionalization can be realized also in this 2d system for temperatures higher than the corresponding 3d transition temperature @xmath21 .
another spectacular experiments on spin-1/2 quasi 2d triangular systems have been recently reported by shimizu _
et al_. @xcite here two organic materials @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 and @xmath1-(et)@xmath2cu@xmath2[n(cn)@xmath2]cl ( where et denotes bis(ethylenedithio)-tetrathiafulvalence ) are synthesized , corresponding essentially to two different values of @xmath26 0.80.9 and @xmath27 , respectively .
@xcite while the latter material shows commensurate spin ordering at @xmath28 k , for the former material no magnetic order is observed even down to the milli - kelvin region ( @xmath29 mk ) regardless the fact that the estimated value of @xmath5 is about @xmath30 .
this observation strongly indicates possible realization of a spin liquid state for @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 .
@xcite motivated by these experiments , we shall consider here the spin-1/2 antiferromagnetic heisenberg model on the triangular lattice with spatial anisotropy described by eq .
( [ model ] ) .
in particular , we report a detailed and systematic quantum monte carlo ( qmc ) study of the ground state as well as the low - lying excitations of this model system as a function of @xmath4 . using both quantum variational monte carlo ( vmc ) method and green function monte carlo method with an improved extension of the fixed node approximation ( named effective hamiltonian approach in this paper )
, it is found that there exist two spin liquid regions in the phase diagram of this model by varying the ratio @xmath4 for @xmath31 .
the first one is stable for small values of the coupling @xmath32 , and appears to show gapless , fractionalized fermionic excitations .
the second one is energetically favored in the region @xmath8 , and is a more conventional spin liquid with a small spin gap in the excitation spectrum , the same type of spin liquid phase realized in the quantum dimer model in the isotropic triangular geometry . @xcite it is argued that the two experimental observations of spin liquid like behaviors for @xmath0 and @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 mentioned above should correspond to these two different spin liquid phases of the model , respectively .
the paper is organized as follows . in sec .
[ method ] , we first introduce the variational wave functions considered ( sec . [ vwf ] ) , and describe our original optimization method to obtain the minimum energy variational wave function containing a large number of variational parameters ( sec .
[ srmethod ] ) . in order to systematically correct this variational ansatz ,
an effective hamiltonian approach is introduced in sec .
[ fnemethod ] along the line of the well established diffusion monte carlo technique , @xcite allowing , for continuous systems , to achieve the best variational wave function with the same phases of the chosen variational state . as explained in sec .
[ secdyn ] , within this approach , it is also possible to calculate the low - lying excitation spectrum , which can be compared directly with dynamical experimental measurements . in sec .
[ secresults ] , all our numerical results are reported for the spin-1/2 antiferromagnetic heisenberg model on the triangular lattice as a function of @xmath4 , including the decoupled chains case ( @xmath14 ) . finally our conclusions and remarks are presented in sec .
[ conclusion ] .
the paper is also supplemented by several important appendices for the detailed explanation of the methods and the wave functions used , which should be very useful for reproducing our results or extending our approach to other model systems .
a part of this work have been reported briefly as a short communication .
the variational wave function considered in this study is the so - called projected bcs state defined by @xmath33{1.25mm}{0.25mm}}{\rm bcs}\rangle}={\cal p}_{\rm g } \exp\left [ \sum_{i < j } f_{i , j } \left(c^{\dag}_{i,\uparrow } c^{\dag}_{j,\downarrow}+ c^{\dag}_{j,\uparrow } c^{\dag}_{i,\downarrow } \right ) \right ] |0\rangle,\ ] ] where @xmath34 ( @xmath35 ) is an electron creation ( annihilation ) operator at site @xmath36 with spin @xmath37(@xmath38 ) , @xmath39 is the vacuum state with no electrons , the function @xmath40 is the so - called pairing function that contains all variational freedom of the @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] , and is determined by the minimum energy condition , whereas @xmath42 is the usual gutzwiller projection operator onto the subspace of singly occupied sites , implying that the total number of electrons @xmath43 is equal to the number of sites @xmath44 .
the pairing function @xmath40 of this projected bcs state can be parameterized using a bcs hamiltonian : @xmath45\ ] ] here @xmath46 and @xmath47 as well as the chemical potential @xmath48 ( assumed uniform ) can be considered variational parameters , which implicitly determine the pairing function @xmath40 corresponding to the ground state ( gs ) of @xmath49 . here
@xmath50 ( @xmath51 ) label the sites of the lattice ( see fig .
[ lattice ] ) , _ i.e. _ , @xmath52 , in the lexicographic order , so that the condition @xmath53 in eq . ( [ wf ] ) is meaningful in any spatial dimension . when @xmath46 and @xmath47 depend only on @xmath54 , _
i.e. _ , @xmath55 and @xmath56 , respectively , @xmath49 can be described more compactly by @xmath57 here @xmath58 @xmath59 and @xmath60 for a singlet pairing , @xmath61 , and thus @xmath62 . in this case the projected bcs wave function defined by eq .
( [ wf ] ) reads @xmath63{1.25mm}{0.25mm}}{\rm bcs}\rangle}={\cal p}_{\rm g } |{\rm bcs}\rangle,\ ] ] with @xmath64 being the ground state of @xmath49 given by eq .
( [ bcshk ] ) : @xmath65 |0\rangle , \label{wfbcs}\ ] ] where @xmath66 , @xmath67}$ ] , and @xmath68 with @xmath69 . at the variational level
, both @xmath70 and @xmath71 have to be parameterized in order to minimize the variational energy of the @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] wave function .
the most relevant parameters for lowering the energy are the short range terms , and we have chosen to expand @xmath72 , where @xmath73 are variational parameters and @xmath74 are the nearest neighbor vectors ( @xmath75 ) .
analogously , the gap function @xmath76 is truncated up to the third nearest distance along the chain ( @xmath77 ) direction .
this is also because , as will be discussed in sec .
[ sub:1d ] , for the 1d spin-1/2 antiferromagnetic heisenberg model @xmath78 , inclusion of the parameter @xmath79 is known to be crucial for this type of projected bcs states to represent almost exactly the ground state of @xmath78 .
@xcite for the present 2d system described by eq .
( [ model ] ) , which preserve @xmath80 symmetry for @xmath81 , the projected bcs state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] thus contains ten independent variational parameters in @xmath71 ( see , _ e.g. _ , tab .
[ tablecoldea ] ) and the chemical potential @xmath82 which , as opposed to the 1d case , may differ from zero in the triangular lattice case . as will be discussed in sec .
[ sub : coldea ] , we found that the variational parameters @xmath83 and @xmath84 are irrelevant and the energetically favorable symmetry of @xmath71 is @xmath85 .
@xcite as shown in sec .
[ secresults ] , the projected bcs state described above is a very good variational state for @xmath86 0.7 .
however , close to the isotropic limit @xmath87 , the translation invariant ansatz state , also previously attempted in the presence of hole doping , @xcite is not very accurate when compared with the exact diagonalization results possible on the @xmath88 cluster .
@xcite in this region of @xmath4 , as shown in sec .
[ secresults : sondhi ] , we have found that it is more convenient to consider a bcs hamiltonian defined on a @xmath89 unit cell [ cf .
( [ pairbroken ] ) ] for a much better variational wave function . as shown in app .
[ appendix : sondhi ] , by using the projected bcs state thereby constructed , it is then possible to represent the well known short range resonant valence bond ( rvb ) wave function , with a particular choice of the variational parameters . the rvb state is a good variational ansatz for the isotropic triangular lattice @xcite and represents a very convenient initial guess for defining , within the present @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] framework , an accurate variational state in the nearly isotropic triangular lattice . in order to evaluate the optimal variational parameters that minimize the energy expectation value , @xmath90 we follow the method , recently introduced for calculations of electronic structure , @xcite which will be described in some detail in the following . according to this method , in order to reach the minimum energy @xmath91 in a stable and efficient way , the logarithmic derivative @xmath92 of the wave function @xmath93 with respect to each variational parameter @xmath94 and/or @xmath47 ( @xmath95 ) has to be evaluated on a given @xmath43-electron configuration @xmath96 of the projected hilbert space with one electron per site , namely , @xmath97 in fact , for infinitesimal changes of these variational parameters , @xmath98 , the corresponding change of the wave function reads @xmath99.\ ] ] in order to simplify the notations , here we introduce the operator @xmath100 corresponding to @xmath92 which is defined by @xmath101 using this operator form , the above equation is more compactly written as @xmath102 |\psi_{\{\alpha_k\}}\rangle\ ] ] valid up to @xmath103 . for the minimization method described below ,
it is important to evaluate numerically the value @xmath92 corresponding to a given real space configuration of electrons @xmath104 satisfying the constraint of no doubly occupied sites . to this purpose
, we have to recall that the variational parameters @xmath47 and @xmath46 are explicitly defined in @xmath49 [ eq .
( [ bcsh ] ) ] , but only implicitly in the wave function itself , defined as the ground state of this unprojected bcs hamiltonian @xmath64 [ eq . ( [ wf ] ) ] .
thus , in order to evaluate the logarithmic derivatives of the wave function @xmath105 with respect to a variational parameter @xmath106 , we apply simple perturbation theory to @xmath49 and calculate the perturbed state @xmath107 , within linear order in @xmath108 .
it is then possible to compute @xmath109 using simple algebra and within the same accuracy @xmath103 . after recasting the calculation by using appropriate matrix - matrix operations ,
the evaluation of @xmath92 is possible in an efficient way , using only @xmath110 operations for each variational parameter @xmath106 .
more details of the method are found in app .
[ appmethod ] .
the minimization method used here is similar to the standard and well known steepest descent method , where the expectation value of the energy @xmath111 is optimized by iteratively changing the parameters @xmath106 ( @xmath112 ) according to the corresponding derivatives of the energy ( generalized forces ) : @xmath113 | \psi_{\{\alpha_k\ } } \rangle \over \langle \psi_{\{\alpha_k\ } } | \psi_{\{\alpha_k\ } } \rangle } + 2 e(\psi_{\{\alpha_k\ } } ) { \langle \psi_{\{\alpha_k\ } } | { \hat o}_k |\psi_{\{\alpha_k\ } } \rangle \over \langle \psi_{\{\alpha_k\ } } | \psi_{\{\alpha_k\ } } \rangle } , \ ] ] namely , @xmath114 where @xmath115 in the standard steepest descent method is determined at each iteration by minimizing the energy expectation value .
@xcite the above expressions for the forces @xmath116 can be computed statistically by appropriately averaging the local energy @xmath117 and @xmath92 over a set of configurations @xmath118 , @xmath119 distributed according to the square of the wave function , @xmath120 , generated with a standard variational monte carlo ( vmc ) scheme , namely , @xmath121 and thus @xmath116 can be computed efficiently once @xmath92 and @xmath122 are evaluated for any sampled configuration .
here we assumed that all quantities are real .
however an extension to the complex case is straightforward .
now , for simplicity , we assume that @xmath115 is positive and small enough .
indeed the variation of the total energy @xmath123 at each step is easily shown to be negative for small enough @xmath115 because in this limit @xmath124 thus the steepest descent method certainly converges to the minimum of the energy when all the forces vanish .
let us now generalize the steepest descent method by slightly modifying the basic iteration ( [ itersd ] ) with a suitably chosen positive definite matrix @xmath125 : @xmath126 again , using the analogy with the steepest descent method , convergence to the energy minimum is reached when the value of @xmath115 is sufficiently small and is kept positive constant for each iteration .
in fact , similarly to the steepest descent method , the energy variation corresponding to a small change of the parameters is : @xmath127 and is always negative for small enough @xmath115 , unless the minimum condition of @xmath128 is reached and the variational parameters no longer change because @xmath129 .
it should be noted here that the steepest descent method is a special case with the matrix @xmath130 ( unit matrix ) .
a more convenient choice for the matrix @xmath131 is given by@xcite @xmath132 this matrix can be efficiently evaluated statistically , and similarly to the forces @xmath116 , can be obtained within a vmc scheme once @xmath92 are computed for a set of configurations @xmath133 , @xmath119 distributed according to the wave function squared @xmath120 , namely , @xmath134 \cdot \left [ o_k ( x_l ) - \bar o_k\right]\ ] ] where @xmath135 and @xmath136 are defined in eq .
( [ defobark ] ) . for whatsoever choice of the @xmath137 configurations @xmath133 ,
this matrix remain positive definite regardless of the statistical noise ( at most has vanishing eigenvalues ) , because it is explicitly written in eq .
( [ defoverlap ] ) as an overlap matrix in @xmath138 between the @xmath139 vectors @xmath140 @xmath141 . in this paper , this generalized method defined by eq .
( [ iterforce ] ) with the matrix @xmath142 given by eq .
( [ defsr ] ) will be called stochastic reconfiguration ( sr ) optimization method .
it should be noted here that other convenient types of positive definite matrix have been recently proposed .
@xcite however for the present purpose the improvement does not appear important .
let us next discuss why the choice eq .
( [ defsr ] ) for the matrix @xmath125 in the sr scheme [ eq .
( [ iterforce ] ) ] is particularly simple and convenient compared to the simplest steepest descent method . for a stable iterative method for the energy optimization , such as the sr method and the steepest descent one
, a basic ingredient is that at each iteration the new set of parameters @xmath143 are determined close enough to the previous set @xmath144 in terms of a prescribed distance .
the fundamental difference between the sr minimization and the standard steepest descent method is simply related to the definition of this distance @xmath145 . within the sr scheme ,
@xmath145 is chosen to be the square distance between the two normalized wave functions corresponding to the two different sets of variational parameters @xmath146 and @xmath147 , _
i.e. _ , @xmath148 the reason to normalize the two wave functions before computing their distance is obvious because , within a vmc scheme , the normalization of the wave function is irrelevant for quantum mechanical averages .
the basic advantage of the sr method is the possibility to work directly with the wave function distance @xmath149 .
in fact , this quantity can be explicitly written in terms of the matrix @xmath125 [ eq , ( [ defsr ] ) ] by substituting eq .
( [ derwf ] ) in its definition eq .
( [ defdistsr ] ) , yielding @xmath150 therefore the most convenient change of the variational parameters is to minimize the functional @xmath151 here @xmath152 is the linear change in the energy @xmath153 , and for @xmath154 the leading term of the expansion in small @xmath155 given in eq .
( [ expdelta ] ) can be used .
@xmath156 is a lagrange multiplier that allows for a stable minimization with small change @xmath154 of the wave function @xmath157 .
then the stationary condition @xmath158 naturally lead to the sr iteration scheme described by eq .
( [ iterforce ] ) with @xmath159 . in a similar manner ,
it is also possible to obtain the standard steepest descent method . in this case
the cartesian metric defined in the @xmath139-dimensional space of the variational parameters is implicitly assumed to distinguish the two sets of variational parameters , _
i.e. _ , the distance here is defined to be @xmath160 the same argument used above to minimize @xmath161 with respect to the variational parameter change @xmath162 will then lead to the standard steepest descent algorithm described by eq .
( [ itersd ] ) . the advantage of the sr method compared with the steepest descent one is now transparent .
sometimes a small change of the variational parameters corresponds to a large change of the wave function , and conversely a large change of the variational parameters can imply only a small change of the wave function .
the sr method takes into account this effect through a better definition of the distance @xmath163 [ eq .
( [ defdistsr ] ) ] . here
a single sr iteration of the sr minimization scheme is summarized as follows : i ) a set of variational parameters @xmath164 is given after the @xmath36-th iteration , ii ) the generalized force @xmath116 [ eqs .
( [ forces ] ) and ( [ defforza ] ) ] and the matrix @xmath125 [ eqs .
( [ defsr ] ) and ( [ defoverlap ] ) ] are calculated statistically using a small variational monte carlo simulation ( bin ) containing typically a few thousand samples ( @xmath165 ) distributed according to the wave function squared @xmath166 , and iii ) a new set of the variational parameters @xmath167 is determined from eq .
( [ iterforce ] ) with a suitable choice of @xmath115 .
after a few hundred ( or sometimes thousand ) iterations needed for equilibration , the iteration described above is further repeated to statistically determine the optimized variational parameters until the desired statistical accuracy is reached .
this method allows for a very accurate determination of the optimized variational parameters with very small statistical uncertainty . in the present study
, all the variational parameters are optimized using this sr minimization method . a sample case study for the sr minimization scheme is presented in fig .
[ plotmin ] for the 1d antiferromagnetic heisenberg model on a @xmath168 ring . as seen in fig .
[ plotmin ] , after the first few hundred iterations needed for equilibration , the variational parameters fluctuate around the stable mean values .
it is interesting to notice in fig .
[ plotmin ] that the variational parameters may continue to change substantially , even after the energy appears to reach its equilibrium value only after the first @xmath169 iterations .
the reason is that a very tiny energy gain , not visible in this plot , is implicitly reached at equilibrium , by satisfying very accurately the euler condition of minimum energy , _
i.e. _ , @xmath128 for @xmath170 .
a stochastic minimization method like the steepest descent method or the sr one , which is based not only on the energy but also on its derivatives ( @xmath116 ) , is therefore much more efficient , especially for statistical methods where energy differences for slight changes of the variational parameters are often very noisy .
having determined the optimized variational parameters @xmath171 as described above , a single standard vmc simulation is performed to calculate various physical quantities for @xmath172 using a bin length much larger than the one used in the sr minimization procedure .
nevertheless , the mentioned euler conditions ( @xmath128 for all @xmath173 ) are usually satisfied within a few standard deviations simply because the statistically averaged variational parameters are much more accurate than the ones obtained at the final sr iteration .
@xcite as is well known , within the variational approach , it is sometimes very difficult to describe long range properties accurately .
this is simply because these long range properties are rather insensitive to the energy , which is instead determined mainly by short range correlation functions .
a clear example of this limitation is found when this method is applied , for instance , to study the long range antiferromagnetic order for the spin-1/2 antiferromagnetic heisenberg model on the square lattice .
the long range antiferromagnetic order is now considered a well accepted and established property of this model .
@xcite nevertheless it has been shown in ref .
that two wave functions with completely different long - distance properties , with or without antiferromagnetic long range order , provide similar and very accurate ( within @xmath174 accuracy ) energy per site in the thermodynamic limit . here
we shall consider a possible way to overcome this limitation of the variational method by introducing what will be named `` effective hamiltonian approach '' in the following .
the main idea of this method is to approximate , as accurately as possible , the exact hamiltonian @xmath15 by an effective hamiltonian @xmath175 , for which the exact ground state can be numerically sampled by green function quantum monte carlo schemes .
the important point is that , within this construction , some of the important short range properties of the hamiltonian @xmath15 are preserved , and therefore a much better control of correlation functions is achieved compared to a mere variational approach .
the short range properties of the hamiltonian @xmath15 can be conveniently defined in the configuration basis @xmath176 .
each element of this basis is defined by the positions and spins of the @xmath43 electrons . in this basis
the matrix elements of the hamiltonian @xmath15 are then indicated by @xmath177 . in this paper , we define a short range hamiltonian as a hamiltonian @xmath15 for which the _ off - diagonal _ matrix elements @xmath178 are non zero only for configurations @xmath104 and @xmath179 connected one to the other by local short - range `` moves '' of electrons ( _ e.g _ , hoppings or spin - flips ) . more precisely , @xmath180 h_{x^\prime , x } = 0 & ~~{\rm otherwise } \end{array } \right.\ ] ] where @xmath181 determines the range of the off - diagonal matrix elements . in this definition ,
not only conventional hubbard , heisenberg , and @xmath182-@xmath5 models can be thought of short range hamiltonians ( with @xmath183 ) , but also models with any interaction @xmath184 which is diagonal in the configuration basis @xmath185 ( @xmath186 ) , _
e.g. _ , the long range coulomb interactions , can be considered short range hamiltonians . therefore , within this definition , most of the physically relevant hamiltonians are short range , essentially because quantum fluctuations are important only through the short range off - diagonal matrix elements , in the absence of which the hamiltonian is simply classical . at present ,
short - range spin hamiltonians which can be solved numerically by quantum monte carlo methods are the ones for which the quantity @xmath187 is strictly negative for all non zero matrix elements @xmath188 .
whenever a wave function @xmath189 satisfying this condition exists , a unitary transformation @xmath190 { \rm sgn } \left[\psi_g(x^\prime)\right ] h_{x , x^\prime } $ ] transforms the hamiltonian to a standard bosonic one ( namely with all non - positive off - diagonal matrix - elements ) that can be solved numerically with quantum monte carlo techniques , without facing the so - called sign problem .
@xcite for instance , for the antiferromagnetic heisenberg model in any bipartite lattice , a particularly simple variational wave function @xmath191 satisfying the marshall sign rule , @xmath192 , allows to satisfy the condition @xmath193 , where @xmath194 is the number of electrons with down spin on one of the sublattices for the given configuration @xmath195 ( see app .
[ app : marshall ] ) . with this wave function @xmath189
, it is therefore possible to solve numerically the spin-1/2 antiferromagnetic heisenberg model in 1d , on a two - leg ladder , and on the 2d square lattice . in these cases , it is also clear that with the same marshall sign of the wave function @xmath189 , different low energy properties can be obtained , _
i.e. _ , a gapless spin liquid ground state for the single chain , a gapped spin liquid for the two - leg ladder , and a quantum antiferromagnet for the 2d limit .
however , as is well known , only for very particular models the marshall sign rule and eq .
( [ nosignproblem ] ) are satisfied , @xcite and in general eq .
( [ nosignproblem ] ) is violated regardless of the wave function @xmath196 used . even when a wave function @xmath196 with the optimal signs , _
i.e. _ , the exact gs signs , is used for generic frustrated hamiltonians , there still exist off - diagonal matrix elements with @xmath197 ( notorious sign problem ) .
namely , due to frustration , they do not decrease the energy expectation value . in order to overcome this difficulty and treat more generic models
, we will define below an effective hamiltonian @xmath175 , which is closely related to @xmath15 , by using an optimal wave function @xmath189 , that we name in the following the guiding function .
this guiding wave function is required to be non zero for all configurations @xmath195 .
even if this requirement is not satisfied , all the forthcoming analysis remain valid in the subspace of configurations @xmath195 for which @xmath198 . once the guiding wave function is provided and thus @xmath175 is defined , the effective model system @xmath175 is solved exactly using the standard green function quantum monte carlo method .
@xcite as will be shown later , the low energy properties of @xmath175 are weakly dependent on the low energy properties , _
i.e. _ , long range behavior , of @xmath189
. the gs wave function of @xmath15 is approximated by the gs of an effective hamiltonian @xmath175 .
this approximate variational state is very good in energy because most of the matrix elements of @xmath15 are treated exactly in @xmath175 , whereas the remaining ones are removed and traced to the diagonal terms of @xmath175 .
as it will be shown later on , this enforces the constraint that the gs of @xmath175 has the same non trivial signs of @xmath189 .
therefore , if @xmath189 is chosen to have the same signs of the exact ground state of @xmath15 for most configurations , this approach becomes essentially exact .
@xcite within the effective hamiltonian approach , the problem to construct an accurate approximate wave function for the gs of @xmath15 is therefore related to how to know phases of the gs .
this , we believe , is a much simpler task , because a good variational wave function of a short range hamiltonian should provide also good phases .
in fact , the variational energy of the guiding function @xmath196 , @xmath199 \left|\psi_g(x ) \psi_g(x^\prime ) h_{x , x^\prime}\right| \over \sum_x
|\psi_g(x)|^2 } , \ ] ] depends strongly on the signs of @xmath189 via the short range terms appearing in @xmath200 . this assumption that a variational wave function with accurate energy should have very good signs can be checked directly on small size clusters . moreover , for the nearest neighbor antiferromagnetic heisenberg model on the square lattice , it is known that all good variational wave functions satisfy the marshall sign rule , although they may display different large distance behaviors .
@xcite this fact clearly confirms this assumption because all these wave functions differ only in their amplitudes but not in their phases . from these considerations , in this paper , we will chose as a guiding function the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] described in the previous subsection with the variational parameters optimized for each @xmath4 by a careful energy minimization using the sr scheme ( sec .
[ srmethod ] ) . in most cases ,
due to the quality of the guiding function used , no relevant corrections are found in the various large distance correlations calculated for @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] when these are compared with the ground state correlations of @xmath175 . in some special cases ,
the use of the effective hamiltonian approach is instead of crucial importance .
\end{array } \right .
\label{fnham}\ ] ] where @xmath205 the fn approach was inspired from the similar fixed node method on continuous systems , @xcite and indeed is a well established approach @xcite which provides also variational upper bounds of the ground state energy , _
i.e. _ , @xmath206 where @xmath207 is the ground state energy of @xmath202 .
as we will show below , for lattice systems there is a better way to choose this effective hamiltonian @xcite , which not only provides better variational energies , but also allows for a better accuracy of the low energy long distance properties of the ground state . for this end
, it is important to notice the following key difference between a lattice system and a continuous one : in the lattice system the configurations @xmath185 which do not satisfy the condition ( [ nosignproblem ] ) may be a relevant fraction of the total number of configurations , whereas in the continuous case such configurations represent just an irrelevant nodal surface of the phase space . therefore dropping all the off - diagonal matrix elements with @xmath208 as in the standard fn method seems to be a too drastic approximation for the lattice case .
this approximation can be indeed improved for lattice systems because , contrary to the continuous case , the fn hamiltonian does not provide the best variational state with the same signs of the chosen guiding function @xmath189 .
whenever there is no sign problem and @xmath215 , @xmath216 is obviously the best choice for which @xmath217 .
also this choice @xmath216 coincides with the standard lattice fn approach [ eq .
( [ fnham ] ) ] ( which will be denoted by the acronymous fn ) .
conversely , when the sign of the off - diagonal term in @xmath15 are frustrated ( @xmath218 ) , a better choice of the constant @xmath214 can be obtained by using a relation which has been well known for continuous systems , and used to correct efficiently the error due to the finite time slice discretization in the diffusion monte carlo ( dmc ) calculations .
@xcite let us first discuss this relation used in dmc before considering the lattice case . in the dmc , the small imaginary time ( @xmath219 ) evolution of the electron configurations
is governed by the exact hamiltonian propagation , @xmath220 , with a diffusion coefficient @xmath221 determined only by the free kinetic operator in @xmath15 .
it is then possible to correct the approximate finite @xmath219 dynamic corresponding to the fixed node hamiltonian , by requiring that it satisfies exactly the short time diffusion condition .
this condition can be simply written as @xmath222 ] = d,\ ] ] where @xmath223 is the diffusion coefficient in three dimensions , and @xmath224 is the electron position operator . for lattice systems with periodic boundary conditions ( pbc ) ,
the position operator @xmath224 is not well defined , as it can not be matched with the boundary conditions .
this limitation can be easily solved by using periodic spin position operators defined in the exponential form by @xmath225 \end{aligned}\ ] ] where @xmath226 is the @xmath227-component of the spin operator at site @xmath228 , and @xmath229 labels the spatial coordinates , _
e.g. _ , @xmath230 and @xmath231 for a @xmath232 square lattice .
these operators are diagonal in the basis of configurations @xmath185 , @xmath233 , as is the analogous position operator @xmath224 in the continuous case .
remarkably @xmath234 is exactly equivalent to the well known lieb - schultz - mattis operator , used to show a well known property on the low energy spectrum of spin-1/2 heisenberg hamiltonians @xcite .
after simple inspection , a relation similar to the one ( [ diffusion ] ) can be found also for the periodic spin position operators @xmath234 , by simply imposing the following equation : @xmath235 \right ] |\psi_g \rangle = \langle \psi_g | \left [ { \hat x}^{\dag}_{\nu } , \left [ { \hat h}^{\rm eff } , { \hat x}_{\nu } \right ] \right ] |\psi_g \rangle\ ] ] for the lattice case , both the left hand side and the right hand side of this relation have non trivial expectation values .
both of them can be simply calculated by a standard vmc method without particular difficulty , after recasting them in a more conventional form for vmc calculations : @xmath236 \right ] |\psi_g \rangle & = & -\sum_x \sum_{x^\prime } \psi_g(x ) \psi_g(x^\prime ) { \cal h}_{x , x^\prime } |x_{\nu}(x)-x_{\nu } ( x^\prime ) |^2
\nonumber \\ & = & -\sum_{x } |\psi_g(x)|^2 \left [ \sum_{x^\prime } \psi_g(x^\prime ) { \cal h}_{x , x^\prime } |x_{\nu}(x)-x_{\nu } ( x^\prime ) |^2/\psi_g(x ) \right ] , \end{aligned}\ ] ] a relation that is obviously valid both for @xmath237 and for @xmath238 .
here we assumed @xmath239 .
therefore , similarly to the continuous scheme@xcite , the value of the constant @xmath214 can be determined from eq .
( [ relation ] ) with high statistical accuracy .
notice that if there is no sign problem , _
i.e. _ , @xmath240 for all configurations @xmath195 , the constant @xmath214 turns out to be exactly one with vanishing statistical error , yielding again @xmath241 . after determining the constant @xmath214 , the effective hamiltonian @xmath175 [ eq . ( [ gammaham ] ) ]
is defined , and the ground state @xmath242 with its eigenvalue @xmath243 and the corresponding low energy excitations of @xmath175 can be computed using the standard green function quantum monte carlo method @xcite without sign problem . to compute the expectation value of the energy @xmath244 using @xmath245 as an approximate ground state for @xmath15 , the method described in ref . can be applied , which very often improves sizably the upper bound estimate of the energy , _
i.e. _ , @xmath246 , even in the standard fn case with @xmath216 .
@xcite as also remarked in ref .
, contrary to the continuous case , for lattice hamiltonians the lowest variational energy @xmath247 does not correspond to @xmath216 in general . in the following
, we will indicate by fne the improved fn effective hamiltonian method [ eq . ( [ gammaham ] ) ] with the constant @xmath248 determined by the condition ( [ relation ] ) , using for @xmath196 the lowest energy variational wave function of the form described in sec .
[ vwf ] . the spin - one excitation spectrum of the effective hamiltonian @xmath175 can be calculated by applying the forward walking technique @xcite used to evaluate the imaginary time evolution of the following quantity @xmath249 where @xmath250 and @xmath242 is the ground state of @xmath175 .
note that the imaginary time propagation in eq .
( [ sqimag ] ) can be evaluated without discretization errors in time @xmath251 , as is common to many other quantum monte carlo techniques , @xcite and also pointed out in ref .
for the present monte carlo scheme . by simple inspection , the spin one excitation energy @xmath252 of @xmath175 for momentum @xmath253 can be calculated by fitting the large imaginary time behavior of @xmath254 $ ] . here
@xmath255 and @xmath256 is the ground state energy of @xmath175 .
thus we fit @xmath257 with @xmath258 for @xmath259 , where @xmath260 , @xmath261 , and @xmath262 are fitting parameters , and @xmath263 is a suitable cutoff time , large enough so that the fitting form ( [ fit ] ) can be used with good accuracy .
in fact , the present fit is very stable in @xmath251 , and a satisfactory convergence of @xmath262 is obtained even for relatively small @xmath264 . as a typical example , fig .
[ h1d ] ( a ) shows a semi log plot of @xmath265 at @xmath266 for the 1d spin-1/2 antiferromagnetic heisenberg model @xmath78 . it is clearly seen that for a wide region of @xmath267 , @xmath268 is linear so that we can safely estimate the excitation energy @xmath262 with eq .
( [ fit ] ) .
the error bars for @xmath262 can be efficiently evaluated by using the `` bootstrap technique '' , @xcite because the numerator and the denominator of eq .
( [ sqimag ] ) are highly correlated . to demonstrate the accuracy and reliability of the method ,
the calculated @xmath262 for @xmath78 is shown in fig .
[ h1d ] ( b ) .
as seen in the figure , the agreement between our results and the exact values is excellent .
it is worthwhile to emphasize that , within this method , a single monte carlo simulation allows to calculate all the lowest spin one excitation energies for various momenta with no extra effort .
moreover , the method is not restricted to the computation of the spin triplet spectrum alone , but can be easily generalized to arbitrary operators @xmath269 as long as they are diagonal in terms of the chosen basis @xmath185 .
although the high energy excitations are more difficult to calculate because the signal decays much faster in time , the most important low energy spectrum can be accurately determined .
before presenting our numerical results , here we will show an important relation between the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] and the particle - hole symmetry , @xcite which turns out to be crucial to differentiate spin systems defined on the conventional non - frustrated square lattice and the ones defined on the triangular lattice geometry . in both cases each site of the lattice is represented by a vector @xmath270 with @xmath271 and @xmath272 being integer ( see , _ e.g. _ , fig .
[ lattice ] ) . then a particle - hole transformation is defined with @xmath273 or in the reciprocal lattice space with the reciprocal lattice vectors @xmath274 and @xmath275 , this transformation is equivalently defined with @xmath276 where @xmath277 . as shown in app .
[ app : marshall ] , whenever the bcs hamiltonian @xmath49 is invariant under the particle - hole transformation and @xmath278 is real , _
i.e. _ , @xmath279 \delta_{\bf k } = -\delta_{-{\bf k}+{\bf q } } \\[0.1 cm ] \mu=0 , \end{array } \right.\ ] ] the corresponding projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] [ eq .
( [ wfq ] ) ] satisfies the so - called marshall sign rule .
@xcite the marshall sign rule is an exact property for the ground state of the spin-1/2 antifferomagnetic heisenberg models on non frustrated lattices such as the square lattice , and indeed for these models the minimum variational energy is achieved when this rule is satisfied by the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] .
we shall first show the results for the uncoupled chain limit , _
i.e. _ , for the 1d spin-1/2 antiferromagentic heisenberg model with nearest neighbor coupling : @xmath280 several previous studies @xcite have found that the ground state of @xmath78 can be described very accurately by the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] [ eq .
( [ wfq ] ) ] with only the nearest neighbor hopping @xmath281 and the first three neighbors for the gap function @xmath282 @xmath283 . here
the site-@xmath36 is represented by the vector @xmath284 .
notice that the ground state of @xmath78 satisfies the marshall sign rule , and therefore from eq .
( [ restriction ] ) with @xmath285 , @xmath82 and @xmath286 with @xmath287 even have to be identically zero .
it was also pointed out @xcite that the inclusion of the third neighbor gap function @xmath288 is crucial to improve the accuracy of the @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] . as shown in fig .
[ plotmin](a ) , the optimized parameters for @xmath168 are @xmath289 and @xmath290 with the notations of eq .
( [ defdeltak ] ) , _ i.e. _ , @xmath291 . the corresponding variational estimate of the total energy is @xmath292 which is in excellent agreement with the exact value of @xmath293 . for larger clusters
the variational parameters change slightly and smoothly , and the same kind of accuracy is obtained even in the infinite size limit . by simple quadratic ( linear )
extrapolation in @xmath294 for the energy ( @xmath295 ) with data up to @xmath296 sites , we found @xmath297 [ @xmath298 , @xmath299 which compares very well with the well - known exact value @xmath300 .
this result suggests that our variational method is particularly accurate in describing this 1d exactly solvable case , which is precisely our starting point for studying weakly coupled chains with @xmath301 . in the rest of this subsection
, we shall show that also the low - energy excited states can be described by projected bcs states .
as we mentioned above , to satisfy the marhsall sign rule , the optimized variational parameters in @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] for the ground state meet the constraint relation ( [ restriction ] ) with @xmath285 .
therefore , at this momentum @xmath302 , the spin-1/2 bcs excitation spectrum @xmath303 [ eq . ( [ bcsspectrum ] ) ] shows gapless excitations at @xmath304 , in perfect agreement with the exact spinon spectrum of the bethe - ansatz solution .
@xcite since the elementary excitations of the bcs hamiltonian with energy @xmath305 are described by the standard bogoliubov modes , @xcite @xmath306 \gamma_{-k,\downarrow}= v_k c^{\dag}_{k,\uparrow } + u_k c_{-k,\downarrow } , \end{array } \right.\ ] ] the simplest variational state for the spinon at momentum @xmath253 is @xmath307 to see whether this state @xmath308 corresponds to a spinon state , we consider a ring with @xmath309 number of sites @xmath310 and _ z_-component of the total spin @xmath311 .
for this case it is known that a well defined spinon exists only for half of the total brillouin zone ( @xmath312 ) .
@xcite for this branch , as shown in fig .
[ bethe ] , the wave function @xmath308 represents very well a spinon with momentum @xmath253 , as can be verified by the good accuracy in the energy and its small variance @xmath313 ^ 2.\ ] ] the variance @xmath314 is zero for an exact eigenstate , and is small for a very accurate variational state . as seen in fig .
[ bethe ] , this is clearly the case for the momenta @xmath315 . and
@xmath316 ( lower panel ) and the variance @xmath317 and @xmath318 ( upper panel ) of the projected bcs states with a single bcs mode @xmath319 and three bcs modes @xmath320 ( see text ) as a function of momentum @xmath173 .
the model studied is the 1d spin-1/2 antiferromagnetic heisenberg model on a ring with l=31 . for comparison , in the lower panel ,
the exact lowest excitation energy is plotted by crosses , and the bcs spectrum @xmath305 denoted by dashed line is scaled by a factor to match the exact bandwidth .
the remaining lines are guides to the eye . ] for the remaining branch of momenta ( @xmath321 ) , the excitations for @xmath78 are no longer elementary .
@xcite as shown in fig .
[ bethe ] , although the state @xmath308 is formally defined even for those momenta , it represents a poor representation for the exact excited state .
in fact @xmath262 becomes quite off from the exact value , and @xmath314 considerably increases when the momentum crosses the
spinon fermi surface .
remarkably , by projecting a non - elementary excitation of the bcs hamiltonian , namely , @xmath322 with @xmath323 , which mimics the correct 3-spinon eigenstate , we can achieve a good agreement for the spectrum even in the region outside the spinon fermi surface ( see fig .
[ bethe ] ) .
notice that , although the projection @xmath42 is crucial to gain a quantitative agreement for the spectrum , the bcs spectrum @xmath305 already gives a qualitatively correct feature of gapless excitations with finite spinon velocity at the right momentum @xmath324 ( see fig .
[ bethe ] ) . since the state @xmath308 [ eq .
( [ elementary ] ) ] is very accurate only in the momentum region @xmath325 , we conclude that only half of the elementary excitations of the bcs hamiltonian remain to be well defined excitations for the spin hamiltonian @xmath78 after applying the gutzwiller projection @xmath42 . this effect is expected to hold also in a 2d fractionalized phase .
it turns out that the elementary bcs excitations which describe the correct spinons after applying the projection operator @xmath42 can be obtained by adiabatically switching off the gap function @xmath278 , a process that defines naturally a fermi surface , so that fermion quasiparticles can be created ( destroyed ) in unoccupied ( occupied ) states only . in 2d
it is very difficult to confirm directly the above `` selection rules '' of the gutzwiller projection acting on the elementary bcs excitations .
for instance , to study single spinon excitations , one might think of a 2d system on @xmath326 lattice with @xmath327 odd as a natural extention of the 1d system considered above .
however this 2d system should contain many elementary excitations of spinons as it is easily understood by considering the @xmath328 limit , and thus it is not an ideal system for studying single spinon excitations .
nevertheless , it is very important to have reached a very accurate 1d limit , with the correct spin fractionalization , within the present variational approach , because this approach can be easily extended to higher dimensions .
finally , it is worth mentioning that , according to the gauge theory by wen , @xcite the low energy excitations of several 2d spin liquids described by this @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] variational ansatz , can be understood at the mean field level without taking into account the gutzwiller projection , simply because this projection becomes irrelevant for large distance correlations .
this is the case for a `` @xmath329 gapless spin liquid '' . without entering into too much details of this theory ,
we only mention here that a @xmath329 spin liquid can be described by a @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] for which no su(2 ) gauge transformation remaining in the restricted hilbert space with no doubly occupation allows to eliminate the gap function in the corresponding bcs hamiltonian . in such a case , only the gauge transformation @xmath330 leaves the bcs hamiltonian unchanged , a transformation which defines the @xmath329 group toghether with the identity operation .
most of the spin liquids which we will describe in this paper are @xmath329 spin liquids in the triangular lattice geometry and should be stable according to the theory mentioned above .
@xcite in this subsection , as a first step towards the 2d limit , we shall consider the spin-1/2 antiferromagnetic heisenberg model for weakly coupled chains in the triangular lattice geometry with @xmath4 small ( see fig .
[ lattice ] ) .
note that this region is appropriate for the material @xmath0 , where @xmath331 .
@xcite motivated by the great success of the present variational approach in the 1d system discussed in the preceding section , the variational ansatz state which we shall consider here for this 2d model in the region @xmath4 small is a similar projected bcs state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] described by eqs .
( [ bcshk])([bcsspectrum ] ) . here
both @xmath46 and @xmath47 in the bcs hamiltonian [ eq . ( [ bcshk ] ) ] parameterizing @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] are assumed translational invariant , and therefore they depend only on the relative vector @xmath332 ( @xmath333 and @xmath334 : integer ) between the two sites @xmath36 and @xmath335 , _
i.e. _ , @xmath336 and @xmath76 . also the @xmath337 point group symmetry is assumed for the variational parameters , _
e.g. _ , @xmath338 for the @xmath85 symmetry , where @xmath339 ( @xmath340 ) is the reflection operator about the @xmath341 ( @xmath342 ) plane : @xmath343 to optimize the variational wave function , we use the sr optimization method described in sec .
[ srmethod ] , which enables us to determine the optimized variational parameters with very high accuracy . as shown in tab .
[ tablecoldea ] , the optimized @xmath336 is found to be non zero only for the nearest neighbors along the chain direction @xmath77 ( we set @xmath344 ) , and negligible otherwise , whereas the optimized @xmath76 is instead found to be sizable even among different chains ( for instance , @xmath345 , shown in tab .
[ tablecoldea ] ) , displaying a true two dimensional character .
it is also found that the symmetry of the gap function @xmath346 which minimizes the variational energy is the @xmath85 symmetry under the @xmath337 point group .
finite - size corrections to the variational parameters scale as @xmath347 , and the @xmath348 cluster is found large enough to be close to the thermodynamic limit , at least , for not too small values of @xmath4 .
all the variational parameters for the @xmath348 cluster as a function of @xmath349 are tabulated in tab .
[ tablecoldea ] along with the variational energy .
it should be noted that since the marshall sign rule is no longer satisfied for this model , the constraint relation ( [ restriction ] ) does not have to be met by the variational wave function .
for example , as seen in tab .
[ tablecoldea ] , the optimized chemical potential @xmath82 turned out to be different from zero .
[ cols="^,^,^,^,^,^,^,^",options="header " , ] the excitation spectrum @xmath305 of the corresponding bcs hamiltonian are analytically calculated in app .
[ appendix : sondhi ] , from which several very interesting features are observed .
if the gap functions @xmath350 are restricted to the nearest neighbors , the spectrum of the bcs hamiltonian is generically gapless ( provided @xmath351 ) .
once the gap functions are non - zero for longer bonds , a tiny energy gain is obtained , and correspondingly a finite excitation gap in @xmath305 is opened .
however this gap is very small .
in fact , it is found that the low energy bcs spectrum @xmath305 on finite size is almost indistinguishable from a gapless one , and that the lowest excitation gap in @xmath305 is estimated as small as @xmath352 of @xmath353 for @xmath354 where @xmath355 is the maximum excitation energy of @xmath305 .
@xcite the reason for the appearance of a finite excitation gap in @xmath305 is simply understood because the bcs hamiltonian with broken translation symmetry has two sites per unit cell , and thus the bcs spectrum is in general gapped .
it should be emphasized here that , as explained in the previous subsection , after applying the projection @xmath42 onto the gs @xmath64 of this bcs hamiltonian , the translation symmetry of the projected bcs state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] is recovered . therefore
, a quite new , remarkable possibility to form a spin liquid state with a finite spin gap @xcite and with a single spin per unit cell can be easily established here within the present variational framework _ without breaking the translation symmetry_. 0.75 cm at @xmath356 for @xmath354 calculated using the fne method .
the incommensurate momenta @xmath357 as well as the corresponding value for the classical limit @xcite are plotted in the upper panel . for clarity ,
the variational estimates of the structure factor @xmath358 are subtracted from the fne results @xmath359 .
the optimized projected bcs state @xmath360{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] without spin jastrow factor @xmath361 ( see the text ) is used for the variational calculations . the same wave function
@xmath360{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] is used as the guiding function @xmath196 for the fne calculations .
the system sizes used are indicated in the figure .
, title="fig:",width=283 ] let us now discuss the static magnetic properties for @xmath354 . as seen in fig.[sq_jp07 ] , the finite size scaling analysis of the fne results for the spin structure factor clearly indicates a vanishing magnetic order parameter .
we have also found that the fne spin structure factors as well as the fne spin - spin correlation functions do not differ significantly from the ones computed for the variational spin liquid state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] explained above ( tab .
[ tablesondhi2 ] ) .
this should be contrasted to the isotropic case shown in sec .
[ secresults : sondhi ] , where significant differences were observed .
these results strongly suggest that the spin liquid ansatz state described here by the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] is stable within the present approach .
therefore , another spin liquid region different from the one discussed in sec.[sub : coldea ] might exist in the anisotropic triangular lattice with @xmath4 closer to one , the state of which can be described , at least approximately , by this proposed spin liquid wave function . since the bcs hamiltonian is defined with a @xmath362 unit cell [ eq .
( [ pairbroken ] ) ] , it is natural to ask ourselves if a valence bond @xmath363 ( not @xmath364 ) would be stabilized , which should also exhibit a finite spin excitation gap at the expense of a broken translation symmetry . in order to examine
whether the valence bond solid is energetically more favored than the spin liquid state for the present anisotropic antiferromagnetic heisenberg model , it is sufficient to study the translation symmetry of the optimized projected state because the valence bond solid necessarily breaks this symmetry .
this possibility can be easily considered within our approach by using a bcs hamiltonian that is not invariant under the transformation @xmath365 . in eq .
( [ pairbroken ] ) it is assumed that the gap function @xmath366 connecting sites @xmath11 and @xmath367 depends only on @xmath368 up to the sign . to check the possible instability toward the valence bond solid ,
we have released these constraints for @xmath366 but with remaining in the same @xmath369 unit cell ; for example , @xmath370 can be different from @xmath371 , but @xmath372 .
we have then optimized independently the first four @xmath366 s at the shortest distances : @xmath373 \delta_2 = \delta_{\tau_1,\vec 2 \tau_1 } , \\[0.1 cm ] \delta_3 = \delta_{0,\vec \tau_2},~ { \rm and } \\[0.1 cm ] \delta_4 = -\delta_{\tau_1,\tau_1+\vec \tau_2 } , \end{array } \right.\ ] ] whereas all the other ones are kept fixed at the values obtained assuming the translation invariant ansatz ( shown in tab .
[ tablesondhi2 ] ) .
whenever the optimized parameters satisfy @xmath374 and @xmath375 within the statistical errors , the projected bcs state is a spin liquid .
otherwise the optimized state is a valence bond solid simply because this state is no longer translation invariant .
in order to optimize these variational parameters , we have used the sr minimization method described in sec .
[ srmethod ] .
our results are presented in fig .
[ hist ] , where the monte carlo evolution of these four parameters are plotted for @xmath354 and @xmath376 .
it is clearly seen in fig .
[ hist ] that , although the initial values of the parameters are far off form the symmetric condition ( @xmath374 and @xmath375 ) , after a few hundred sr iterations these parameters converge to the symmetric values ( tab .
[ tablesondhi2 ] ) .
the inset of fig .
[ hist ] shows the monte carlo evolution of the corresponding energy as a function of the sr iterations ( each iteration corresponds to a small variational monte carlo simulation with fixed variational parameters ) .
after the first few sr iterations the energy appears to be trapped in a metastable state with a broken symmetry solution , but then after one hundred sr iterations the energy eventually converges to a lower value corresponding to the spin liquid fully symmetric solution .
these results , therefore , strongly indicate that the optimized state is translation invariant , and spontaneous dimerization is very unlikely to occur in this model because it is not stabilized even when the variational ansatz wave function allows to stabilize this kind of order . 0.75 cm ) as a function of sr iterations for the spin-1/2 antiferromagnetic heisenberg model on the anisotropic triangular lattice with @xmath354 and @xmath376 . here
the sr minimization method explained in sec .
[ srmethod ] is used with @xmath377 [ eq .
( [ iterforce ] ) ] .
inset : the corresponding energy evolution as a function of sr iterations . at each sr iteration ,
the energy is computed for the wave function with the fixed variational parameters given at the corresponding iteration in the main figure .
, title="fig:",width=257 ] in principle , as shown in ref . for 1d systems ,
also a translation invariant state can give rise to a spontaneously dimerized order after applying the projection operator @xmath42 on this state . in order to rule out this possible order ,
we have also calculated explicitly the dimer - dimer correlation functions @xmath378 defined in eq .
( [ def_dim ] ) for the variational wave function described above ( see also tab .
[ tablesondhi2 ] ) .
the results are compared with the ones computed using the fne method .
a typical example of the results is presented in fig .
[ dim_jp07 ] for @xmath354 and @xmath379 . @xcite as seen in fig .
[ dim_jp07 ] , the two results do not show significant differences , confirming that the spin liquid variational ansatz appears very stable in this case .
it should be emphasized once again that the situation is very different from the isotropic case where a magnetic instability was clearly detected with the fne method ( sec .
[ secresults : sondhi ] ) . 0.75 cm
the numerical calculations presented here strongly suggest that a new type of spin liquid discussed here is an appropriate ground state description of the spin-1/2 antiferromagnetic heisenberg model on the anisotropic triangular lattice , at least in the region around @xmath3800.8 ( also see sec .
[ conclusion ] ) . in the present work ,
we have not attempted to determine the critical value of @xmath6 above which the incommensurate magnetically ordered state is stable , which was previously suggested in , _
e.g. _ , ref . .
indeed , there exists another more interesting phase boundary between the two possible spin liquid states , the one presented here and the one considered in sec .
[ sub : coldea ] , which will be discussed in the next section .
in this paper , using various quantum monte carlo techniques , we have studied the ground state phase diagram as well as the low - lying spin excitations for the spin-1/2 antiferromagnetic heisenberg model on the triangular lattice as a function of the spatially anisotropic coupling @xmath4 ( fig .
[ lattice ] ) . we have found numerical evidence for the presence of two different spin liquid states .
the first spin liquid ( `` algebraic spin liquid '' ) is stable for @xmath381 ( see fig .
[ phaseenergy ] ) , and is characterized by gapless spin excitations , thus very similar to 1d spin liquids .
conversely , the other spin liquid is rather a new type of spin liquid state , stable for @xmath382 , and should show a small spin excitation gap . starting from the isotropic limit @xmath383 where the ground state is magnetically ordered ( classical nel ordered ) , quantum fluctuations increase strongly with decreasing the coupling @xmath4 down to zero . therefore , the stability of a spin liquid in this region of the phase diagram is quite clear . the critical coupling @xmath384 discriminating these two spin liquid phases can be determined in principle by comparing the corresponding energy , and our best estimates of the energy as a function of @xmath4 are summarized in fig .
[ phaseenergy ] . from these results
, it is concluded that the critical coupling @xmath385 is about 0.65 . because of the limitation of currently available cluster sizes , our approach can not either determine precisely the critical coupling or describe accurately the nature of the transition . at the variational level
, a first order transition occurs at a critical point where the energy curves of the two spin liquid phases ( as a function of @xmath4 ) , shown in fig .
[ phaseenergy ] , intersect with different slopes . in the same figure , it is also clear that the two slopes become very close within the fne calculations , suggesting that the transition would eventually turn to a conventional second order transition , when the quantum fluctuations are more accurately taken into account .
( denoted by `` 1d - type '' ) and the other described in sec .
[ secresults : sondhi ] ( denoted by `` 2d - type '' ) .
these wave functions are used as the guiding functions for the fne calculations . ]
likewise , we have not tried to determine the critical coupling and to study the nature of the transition between the spin liquid phase and the magnetically ordered phase with incommensurate magnetic order .
the latter phase should appear somewhere around @xmath386 .
@xcite however , it is very difficult to perform a finite size scaling analysis to determine the magnetically ordered incommensurate phase simply because too large clusters are necessary for an accurate estimate of the magnetic moment .
our numerical results were obtained using the quantum variational monte carlo method as well as the lattice fn and fne methods , which are essentially the green function quantum monte carlo method with the fixed node approximation .
the fne ( `` effective hamiltonian approach '' ) is a further improved version of the standard fn method and developed here in sec .
[ fnemethod ] .
although our results might be affected by this approximation in general , we have shown that the present methods provide very sensible results for the isotropic triangular antiferromagnet with @xmath383 ( sec .
[ secresults : sondhi ] ) , and the numerically exact results for the strongly anisotropic limit of the 1d uncoupled chains with @xmath14 ( sec .
[ sub:1d ] ) .
therefore , we have obtained rather reliable results for both limits of the phase diagram , and the same numerical tools have been applied also to the still controversial region of the phase diagram for @xmath387 .
the quality of our approximation in the fn and fne methods depends mostly on the choice of the guiding function @xmath196 , for which we used the best variational ansatz state , optimized using the sr minimization method ( sec .
[ srmethod ] ) .
the optimization of @xmath196 is a crucial ingredient of our approach , because the approximate ground state which we consider , _
i.e. _ the numerically exact ground state @xmath242 of the effective hamiltonian @xmath175 , can be computed with the restriction to have the same signs of @xmath196 .
indeed , on small clusters for which numerically exact diagonalization of the systems can be done , the variational ansatz state described by an projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] provides a very good average sign @xmath388 [ eq . ( [ sign_def ] ) ] for both frustrated and non frustrated systems as discussed in sec .
[ secresults : sondhi ] and also in refs . , , and .
within this approach , which was shown quite reliable , we have found a surprisingly stable spin liquid ( `` algebraic spin liquid '' ) phase in the regime of large anisotropic couplings @xmath389 ( sec . [
sub : coldea ] ) .
our numerical calculations also indicate that this spin liquid state shows gapless , fractionalized spin excitations ( fig .
[ bz_jp033 ] and fig .
[ bz_jp05 ] ) .
@xcite therefore , we predict that this 2d algebraic spin liquid state should show peculiar low energy properties similar to the 1d systems . although our conclusion is based on numerical calculations for rather large clusters ( up to @xmath390 sites ) , to be fair , we can not rule out a very weak instability toward symmetry - broken ordered states , as predicted in the @xmath391 limit by using the susceptibility criterion based on a random phase approximation ( rpa ) .
@xcite even in that study , the instability occurs in an irrelevantly small temperature region , as also pointed out by the authors .
@xcite moreover , it is not clear how reliable the rpa calculation is for finite values of @xmath4 .
we have also found another type of spin liquid phase in the region @xmath4 close to the isotropic limit , _
i.e. _ , for @xmath382 ( sec .
[ secresults : liquid ] ) .
this rather new type of spin liquid state is characterized by a small spin excitation gap , and is described by the projected bcs state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] with a gap function defined by eq .
( [ pairbroken ] ) with a @xmath362 unit cell .
this spin liquid state is an extension of the conventional short range rvb state @xmath392 [ eq .
( [ srrvb ] ) ] , which has been considered before in the context of the isotropic triangular lattice @xcite and is indeed a very good representation of the exact ground state of the isotropic triangular antiferromagnet for small clusters ( sec .
[ secresults : sondhi ] and ref . ) . to extend the short range rvb state ,
we have constructed an exact mapping between the short range rvb state and the projected bcs state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] ( sec .
[ secresults : sondhi ] and app .
[ appendix : pfaffian ] ) . in doing so
, the unit cell of the bcs hamiltonian , defining the @xmath64 state , is expanded to a ( @xmath393 ) unit cell .
this mapping is crucial in the present study because within this approach the short range rvb state can be easily extended and improved systematically by including hopping integrals , a finite chemical potential , and most importantly long range gap functions in the bcs hamiltonian , with no particular numerical effort .
an additional advantage of using the @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] representation is that it is easy to gain qualitative insight of the low - lying spin excitations from the corresponding bcs excitation spectrum @xmath305 .
@xcite for instance , @xmath305 of the bcs hamiltonian , corresponding to the short range rvb state , shows a finite excitation gap simply because @xmath394 ( sec .
[ secresults : sondhi ] and app .
[ appendix : sondhi ] ) , and therefore a finite spin gap is expected in the short range rvb state .
this is indeed the correct property of the short range rvb state , because the presence of a finite spin gap and a very short correlation length have been established before .
@xcite it is worth mentioning a further remarkable property of this new type of spin liquid state described by this projected bcs wave function . without the projection @xmath42 , the bcs state @xmath64 breaks translation and reflection symmetries because the bcs hamiltonian is defined with a @xmath362 unit cell . as a consequence ,
the bcs excitation spectrum @xmath305 has a finite gap in general .
however , as discussed in sec .
[ secresults : sondhi ] , the translation symmetry is recovered after applying the projection operator onto @xmath64 .
therefore , within the projected bcs wave functions , we have discovered a peculiar way to open a finite spin gap without breaking the translation symmetry of the state .
it is also important to emphasize that the reflection symmetry of the projected state is also restored in the thermodynamic limit , as we have systematically tested numerically ( fig .
[ notyet ] ) , which however can not be proved rigorously except for the limiting case of the symmetric short range rvb state .
our finding of stable spin liquid phases in the spin-1/2 anisotropic triangular antiferromagnet is in good agreement with recent studies of the half - filled hubbard model on the triangular lattice with spatial anisotropic hopping based on the gutzwiller approximation @xcite and the variational monte carlo simulations .
@xcite however , it should be remarked here that the gap function in @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] which we found energetically favorable for the nearly isotropic case is rather different from the ones considered in the previous studies .
@xcite our spin liquid state is defined by a bcs hamiltonian with a non translation invariant gap function , whereas the conventional ansatz state such as the ones considered in refs . and
is defined with a homogeneous gap function , which we found much less accurate close to the isotropic point ( see fig .
[ phaseenergy ] ) .
although both the hubbard model in the limit of large on - site repulsion @xmath395 and the heisenberg model considered here should be the same , we just note that , with a variational method sensitive only to the energy , it is much easier to find good variational wave functions for a low energy effective heisenberg model rather than for the corresponding hubbard model , because the latter contains also the large energy scale @xmath395 . in order to have better insight on the low - lying spin excitations of the spin liquid phases
, we have directly calculated the spin one excitation dispersion with the method described in sec .
[ secdyn ] .
this quantity is particularly important since it can be compared directly with inelastic neutron scattering experiments .
the detailed measurements on @xmath0 by coldea _ et al _ @xcite are indeed available , for which a spin liquid like behavior has been observed .
@xcite it has been also reported that this material can be described by the model hamiltonian eq .
( [ model ] ) with @xmath396 .
@xcite as shown in fig .
[ sqw ] , our calculations were found to be in excellent agreement with the experiments with no fitting parameters .
it is also clear from fig .
[ sqw ] that the dispersion is 2d characteristic because for uncoupled chains ( @xmath14 ) the dispersion should be symmetric around the momenta @xmath397 and @xmath398 .
this excellent agreement strongly supports the exitance of this type of 2d spin liquid state .
the successful comparison also indicates that our numerical technique can compute accurately the excitation dispersion for the non - trivial strongly anisotropic antiferromagnetic heisenberg model on the triangular lattice . encouraged by these results
, we have also calculated the low - lying spin one dispersion for the other spin liquid phase which appears in the region of @xmath8 . a typical results for @xmath399
is reported in fig .
[ sqw_jp08 ] .
this coupling regime is relevant for the organic material @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 for which a spin liquid like behavior has been also observed .
@xcite since there is no experimental data available for the spin excitations of this material , our results shown in fig .
[ sqw_jp08 ] provide a theoretical prediction for the spin excitation dispersion , which should be compared with the future neutron scattering experiments .
it is interesting to notice that the dispersions for @xmath399 and for @xmath400 are very similar along the @xmath104-direction , whereas the difference becomes evident in other momentum regions , for example , along the @xmath401-direction .
this feature should be also checked experimentally in the feature .
vskip 0.5 cm -2.0 cm in both spin liquid phases described by our variational ansatz , the spin excitation gap in the thermodynamic limit can not be resolved by directly computing the excitation dispersions with the method introduced in sec .
[ secdyn ] as the present available system sizes are not large enough . according to the argument presented in the previous section , for @xmath402 a finite spin excitation gap is implied by the corresponding gap in the bcs excitation spectrum @xmath305 of the @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] ansatz .
however , this gap should be very small because i ) it is close to a transition to an ordered state appearing for larger @xmath4 and ii ) it is due to the rather small long range part of the gap functions @xmath47 s . in fact , we have found that the lowest excitation gap in @xmath305 is as small as @xmath403 of @xmath353 for @xmath399 ( where @xmath355 is the maximum excitation energy in @xmath305 ) .
@xcite this small gap might also explain an apparent finite spin susceptibility observed on @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 , @xcite which should eventually vanish exponentially at very low temperatures with an activated behavior .
it is also very interesting to note that the organic material @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 , which shows a spin liquid like behavior under ambient pressure,@xcite has been very recently found to become a superconductor under small applied pressure ( about 46 @xmath404 gpa ) .
@xcite if we assume that the origin of the superconductivity is intimately related to the spin liquid like nature of the phase observed next to the superconducting phase , the present study implies a rather unique , unexpected pairing symmetry of the superconductor with a non translation invariant pairing amplitude . this is a measurable effect because the projected bcs state @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] discussed in sec .
[ secresults : liquid ] is no longer translation invariant once the projection constraint @xmath42 is released or mobile carriers are introduced into the homogeneous spin liquid state .
therefore , we expect that this unconventional pairing formation can be probably observed by , _ e.g. _ , scanning tunneling microscopy experiments .
@xcite based on our numerical calculations and their comparison with experiments , we conclude that the spin liquid is a generic phase of quantum matter , and that , for its stability , it is not important to be very close to a phase transition , unlike the recently proposed scenario in which a spin liquid state appears only at a transition @xmath405 between a magnetically ordered phase and a spontaneously dimerized phase .
@xcite within the projected bcs wave function approach , spontaneous dimerization can be correctly described in quasi 1d systems , but not in 2d unless the translation symmetry of the variational state is explicitly broken .
@xcite even allowing this possible symmetry breaking , by adding appropriate symmetry breaking terms to the bcs hamiltonian , we have not found a stable dimerized phase in the present 2d system .
instead , we have found rather stable spin liquid phases .
finally , our phase diagram of the spin-1/2 antiferromagnetic heisenberg model on the triangular lattice is summarized in fig .
[ phase ] , where possibly related materials are also indicated
. in particular , it would be very important to distinguish experimentally the two different spin liquid phases predicted here for different values of the coupling @xmath406 .
( @xmath6 ) is the nearest neighbor antiferromagnetic coupling between the spins in the chain ( in different chains ) ( see fig .
[ lattice ] ) .
sl and afmlo stand for spin liquid and antiferromagnetic long range order ( incommensurate spin order for @xmath4 away from the isotropic case ) , respectively .
materials possibly described by this class of models are also indicated .
, width=321 ] we thank f. becca and l. arrachea for providing us exact diagonalization results for the @xmath88 clusters , and especially r. coldea for sending us his neutron scattering data on @xmath0 .
one of us ( s.s . ) acknowledges kavli institute for theoretical physics ( kitp ) in santa barbara for its kind hospitality at the early stage of this work . this work has been partly supported by infm and miur ( cofin 2004 ) .
in this appendix , we will explain how to define appropriately and calculate efficiently the quantity @xmath92 [ eq .
( [ opx ] ) ] for each variational parameter @xmath106 of the wave function @xmath105 .
since the derivation is rather general and not restricted to the particular form of the wave function used , we will here indicate the determinantal part of the wave function by @xmath407 , instead of @xmath64 , and thus @xmath408 since it is generally found much more efficient , we will consider a wave function with definite number @xmath43 of electrons , for which there are also important simplifications for deriving computationally convenient expressions for the quantities @xmath92 . to this purpose
, we will use a particle - hole transformation introduced by yokoyama and shiba , @xcite which is defined by the following canonical transformation : @xmath409 c^\dag_{i{\downarrow}}=(-1)^{i_1+i_2 } { \tilde{c}}_{i+l}. \end{array } \right.\ ] ] here @xmath410 ( @xmath411 ) is an electron creation ( annihilation ) operator at site @xmath412 with spin @xmath413 , whereas @xmath414 and @xmath415 represent the new canonical operators corresponding to the original spin up ( spin down ) electrons for @xmath416 ( @xmath417 ) .
@xmath44 is the total number of sites .
after this particle - hole transformation ( [ spindownph ] ) , it is clear that the bcs hamiltonian [ eq . ( [ bcsh ] ) ] can be written as @xmath418 where @xmath419 are appropriate @xmath420 matrix elements , which can be straightforwardly computed from eq .
( [ bcsh ] ) .
note that this matrix can be easily evaluated also for rather general hartree - fock , bcs hamiltonians containing , _
e.g. _ , non translation invariant terms , or several types of orders
. the total number of particles @xmath421 after the transformation ( [ spindownph ] ) is related to the total spin along the @xmath227-direction @xmath422 in the original representation through the following equation : @xmath423 since @xmath424 is conserved in the original bcs hamiltonian , the transformed hamiltonian ( [ hamg ] ) has a definite number @xmath421 of particles , as anticipated . with this particle - hole transformation ,
it is possible to study the spin excitations of the bcs hamiltonian with @xmath425 , as well as the ground state which belongs to the singlet @xmath426 sector .
therefore , in the following , we will consider the general case with unrestricted @xmath421 . by using an appropriate unitary transformation , @xmath427 \displaystyle{{\hat\gamma}^\dag_\alpha=\sum_{i = i}^{2l}\left(\bar{u}\right)_{i\alpha } } { \tilde{c}^\dag}_i , \end{array } \right.\end{aligned}\ ] ] @xmath428 is diagonalized with a new set of quasiparticle operators @xmath429 : @xmath430 where , for convenience , the eigenvalues are sorted in ascending order @xmath431 .
a natural choice of the slater determinant part @xmath432 of the variational wave function is the ground state of @xmath433 with @xmath421 particles : @xmath434 where @xmath435 is the vacuum ( @xmath436 ) of the hilbert space after the particle - hole transformation .
thus the slater determinant for a particle - hole transformed configuration @xmath437 is @xmath438,\ ] ] where @xmath439 is a @xmath440 matrix , whose elements are @xmath441 @xmath442 s ( @xmath443 ) are the `` positions '' of the @xmath421 particles ( @xmath444 ) corresponding to the configuration @xmath445 , and @xmath446 .
now let us consider small changes for @xmath447 .
since @xmath433 depends linearly on @xmath144 , the perturbed system is described by @xmath448 where @xmath449 is a suitable operator proportional to the chosen variational parameter @xmath106 .
for instance , when the chemical potential is changed @xmath450 , namely a term @xmath451 is added to @xmath49 , the corresponding matrix element @xmath452 is @xmath453 for @xmath454 and @xmath455 for @xmath456 , due to the particle - hole transformation ( [ spindownph ] ) .
here @xmath457 is the kronecker @xmath458-function .
when a small change in a pairing term @xmath459 $ ] is considered , the corresponding matrix element @xmath452 is @xmath460 $ ] . with the basis set formed by the quasiparticle operators @xmath429 , the matrix @xmath461 is transformed as @xmath462 , and thus the first order correction to the state @xmath432 is easily computed as follows : @xmath463 |{\bf\phi}_0\rangle+ { \cal o}(\delta\alpha_k^2)\ ] ] where @xmath464_{\eta\nu}}{{\varepsilon}_\nu-{\varepsilon}_\eta } } } & : { \rm for}~ \eta>{n^{(\rm ph)}}{\rm ~and~ } \nu\le { n^{(\rm ph)}}\\[0.4 cm ] 0 & : { \rm otherwise } \end{array } \right.\ ] ] this expression is well defined as long as @xmath432 is non degenerate , so that the denominator in the definition of @xmath465 is always non zero .
this condition is rather generally satisfied for the bcs hamiltonian .
we can now readily express the state @xmath466 in terms of the @xmath467 basis set , @xmath468|{\bf\phi}_0\rangle + { \cal o}(\delta\alpha_k^2)\ ] ] where @xmath469 . at each iteration of the sr minimization procedure described in sec [ srmethod ]
, @xmath470 has to be computed , because the operators @xmath429 change every time a new set of @xmath144 is calculated .
this can be done using four matrix - matrix multiplications . from eq .
( [ wf1o ] ) , it is now easy to evaluate the perturbed slater determinant @xmath471 , and thus an explicit expression for @xmath92 defined by eqs .
( [ opx ] ) and ( [ opx2 ] ) : @xmath472 here @xmath473 is a local single - particle `` green s function '' , @xmath474 which is computed and updated during the variational monte carlo iterations . since the matrix @xmath470 does not depend on the configuration @xmath475 and @xmath473
is always known , only about @xmath476 operations are required to evaluate @xmath92 for each @xmath173 and for each sampled configuration @xmath475 .
finally , we note that when the variational wave function @xmath105 contains the spin jastrow factor @xmath477 , $ ] _ i.e. _ , @xmath478 , the calculation of @xmath92 corresponding to the variational parameter @xmath479 is much simpler and straightforward , because @xmath92 is simply @xmath480 in this case .
in this appendix , we will show that a bcs wave function projected onto the subspace of singly occupied sites satisfies the marshall sign rule provided the corresponding bcs hamiltonian @xmath481\ ] ] satisfies the following conditions : @xmath482 here @xmath410 ( @xmath411 ) is an electron creation ( annihilation ) operator at site @xmath483 with spin @xmath413 , and @xmath484 and @xmath485 are assumed symmetric under @xmath486 interchange and real .
it is also assumed that the ground state of the bcs hamiltonian @xmath64 is unique , namely with a finite size gap to the first excitation , a condition which can be generally met for non trivial values of @xmath47 and @xmath46 . in what follows
, we will show that the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}={\cal p}_{\rm g } |{\rm bcs}\rangle$ ] constructed from the bcs state above satisfies the marshall sign rule , namely , @xmath487 = ( -1)^{n_{\rm a}},\ ] ] where @xmath194 is the number of down spins on one of the two sublattices , and @xmath475 is an electron configuration with no doubly occupied sites , which can be written as @xmath488 with the @xmath489 factors being ordered from left to right according to the increasing index @xmath36 ( @xmath490 is the spin of the electron on site @xmath36 ) . here
it is assumed that the number of sites ( @xmath44 ) is even .
first , it is convenient to perform the following particle - hole transformation for the down spins only : @xmath491 with this transformation , the bcs hamiltonian is transformed to a standard one - body hamiltonian commuting with the total number of particles , _
i.e. _ , @xmath492,\end{aligned}\ ] ] where @xmath36 denotes the imaginary unit , @xmath493 is the @xmath401-component of the spin matrix , and condition ( [ relationapp ] ) is used . on the other hand , after this transformation ( [ phdown ] ) , the basis of no doubly occupied sites turns into @xmath494 are now defined by the @xmath495 positions of the doubly occupied sites : @xmath496 where @xmath497 . because of the additional phase in eq .
( [ phdown ] ) , @xmath498 is related to @xmath499 by @xmath500 . therefore , using the definition of the marshall sign ( [ marshall ] ) , a state with the marshall sign rule in the original basis @xmath185 is equivalent to a bosonic state in the particle - hole transformed basis @xmath501 : for each configuration @xmath502 , the wave function should be always positive ( or always negative ) . in order to show this bosonic rule for the projected bcs state considered
, it should be noticed that the new basis @xmath501 is invariant under a global rotation of the spins which transforms the spin matrix @xmath503 into @xmath504 : @xmath505 d_{j\downarrow } \end{array } \right ) = { \frac{1}{\sqrt{2}}}\left ( \begin{array}{cc } 1 & i\\[0.1 cm ] -i & -1 \end{array } \right ) \left ( \begin{array}{c } a_{j\uparrow}\\[0.1 cm ] a_{j\downarrow } \end{array } \right).\ ] ] indeed , @xmath506 is transformed into @xmath507 .
it is also easily shown that this transformation ( [ su2 t ] ) factorizes the hamiltonian ( [ bcstransf ] ) as @xmath508,\ ] ] where @xmath509 , @xmath510 and @xmath511 being appropriate one - body hamiltonian matrices whose eigenstates are @xmath512 and @xmath513 , respectively ( @xmath514 ) .
therefore , the ground state @xmath515 of @xmath49 once computed in the basis of doubly occupied and empty sites @xmath501 reads @xmath516 where @xmath439 is a @xmath517 matrix with @xmath518 .
this proves the statement given at the beginning of this appendix .
finally , we note that the condition ( [ relationapp ] ) is satisfied whenever the bcs hamiltonian ( [ bcsgeneral ] ) is invariant under a particle - hole transformation : @xmath519 provided @xmath520 and @xmath485 is real .
( [ restriction ] ) in sec .
[ pbcsphs ] follows readily from this particle - hole symmetric condition .
in this appendix , several important properties are discussed on the ground state @xmath64 of the bcs hamiltonian which breaks primitive lattice translation symmetry by extending the unit cell to @xmath362 .
the resulting state @xmath64 is used to build a projected bcs states @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}={\cal p}_{\rm g}|{\rm bcs}\rangle$ ] for the isotropic and nearly isotropic triangular systems discussed in sec .
[ secresults : sondhi ] and sec .
[ secresults : liquid ] , respectively .
let us start with defining the bcs hamiltonian @xmath521 introduced in sec .
[ vwf ] in a slightly different way .
a part of @xmath521 which is invariant under any primitive lattice translations on the triangular lattice ( fig . [ lattice ] ) can be generally described by the following hamiltonian : @xmath522 \nonumber\\ & + & \sum_{\vec r } \left [ { \sum_{{\vec t}_m}}^{\prime } \delta_{{\vec t}_m } \left(c^\dag_{{\vec r}{\uparrow } } c^\dag_{{\vec r}+{\vec t}_m{\downarrow } } -c^\dag_{{\vec r}{\downarrow } } c^{\dag}_{{\vec r}+{\vec t}_m{\uparrow } } \right ) + { \rm h.c.}\right]\nonumber\\ & -&\mu \sum_{{\vec r},\sigma } c^\dag_{{\vec r}\sigma } c_{{\vec r}\sigma},\end{aligned}\ ] ] where the first sum @xmath523 runs over all lattice vectors @xmath524 ( @xmath525 : integer ) , whereas the second sum @xmath526 is for @xmath527 ( @xmath528 : integer ) with @xmath529 or with @xmath530 and @xmath531 as denoted by solid circles in fig .
[ lattice2 ] ( a ) .
( denoted by solid circles ) considered for the sum over @xmath527 .
each circle corresponds to a pair of integers @xmath532 .
( b ) : the first brillouin zone ( hexagon drown by dashed lines ) for the triangular lattice shown in fig . [ lattice ] .
the corresponding reciprocal lattice vectors are @xmath533 and @xmath534 . the reduced brillouin zone ( tilted rectangle )
is also given by solid lines .
several symmetric points are @xmath535 : ( 0,0 ) , p : @xmath536 , q : @xmath537 , p : @xmath538 , and q : @xmath539 . ]
now let us add to @xmath540 an additional pairing term @xmath541 which breaks the underlying lattice translation symmetry : @xmath542 . \end{aligned}\ ] ] note that the unit cell of the total bcs hamiltonian @xmath543 is now extended to @xmath362 . in the following , @xmath544 , @xmath545 , @xmath546 , and @xmath82
are all assumed to be real .
let us first explore the excitation spectrum of @xmath521 .
for this purpose , it is convenient to fourier transform the bcs hamiltonian to the momentum space with the reciprocal lattice vectors , @xmath533 and @xmath534 [ see fig .
[ lattice2 ] ( b ) ] . after the fourier transformation with @xmath547 ,
@xmath49 can be conveniently described by the following @xmath548 matrix form : @xmath549 0 & \xi_2({\vec k } ) & { \delta_{12}({\vec k})}^ * & \delta_{22}({\vec k } ) \\[0.1 cm ] \delta_{11}({\vec k } ) & \delta_{12}({\vec k } ) & -\xi_1({\vec k } ) & 0 \\[0.1 cm ] { \delta_{12}({\vec k})}^ * & \delta_{22}({\vec k } ) & 0 & -\xi_2({\vec k } ) \end{array } \right ) \left ( \begin{array}{c } c_{{\vec k}{\uparrow } } \\[0.1 cm ] c_{{\vec k}+{\vec q}{\uparrow } } \\[0.1 cm ] c^\dag_{-{\vec k}{\downarrow}}\\[0.1 cm ]
c^\dag_{-({\vec k}+{\vec q}){\downarrow } } \end{array } \right ) , \label{mat_broken}\end{aligned}\ ] ] where @xmath550 \displaystyle { \xi_2({\vec k})=-2{\sum_{{\vec t}_m}}^{\prime}t_{{\vec t}_m } \cos\left[\left({\vec k}+{\vec q}\right)\cdot{\vec t}_m\right ] - \mu } , \end{array } \right.\ ] ] and @xmath551 \displaystyle { \delta_{22}({\vec k})=2{\sum_{{\vec t}_m}}^{\prime}\delta_{{\vec t}_m } \cos\left[\left({\vec k}+{\vec q}\right)\cdot{\vec t}_m\right]}\\[0.2 cm ] \displaystyle { \delta_{12}({\vec k})={\sum_{{\vec t}_m}}^{\prime } { \bar{\delta}}_{{\vec t}_m}\left ( { \rm e}^{i\left({\vec k}+{\vec q}\right)\cdot{\vec t}_m } + { \rm e}^{-i{\vec k}\cdot{\vec t}_m } \right ) } \end{array } \right.\ ] ] with @xmath552 so that @xmath553 , the primed sum @xmath554 runs over the reduced brillouin zone shown in fig . [ lattice2 ] ( b ) , or equivalently , _
e.g. _ , for parallelogram lattice of @xmath555 ( @xmath556 , @xmath557 : even ) with @xmath558 the primed sum is taken over @xmath559 and @xmath560 .
the excitation spectrum of the bcs bogoliubov mode for this bcs hamiltonian is now easily calculated by diagonalizing the @xmath548 hamiltonian matrix given in eq .
( [ mat_broken ] ) .
it turned out that the excitation spectrum @xmath561 is doubly degenerate at each momentum @xmath562 and is given by @xmath563 , where @xmath564^{1/2}\ ] ] and @xmath565\nonumber\\ q_1({\vec k } ) & = & { \frac{1}{2 } } \left [ \left ( \xi_1({\vec k})^2 - \xi_2({\vec k})^2 + \delta_{11}({\vec k})^2 - \delta_{22}({\vec k})^2 \right)^2 + 4\left|\delta_{12}({\vec k})\right|^2\left\{\left(\xi_1({\vec k})-\xi_2({\vec k})\right)^2 + \left(\delta_{11}({\vec k})+\delta_{22}({\vec k})\right)^2 \right\}\right]^{1/2}. \nonumber\end{aligned}\ ] ] from this expression , it is clear that generally @xmath566 has a finite excitation gap except for some special cases .
so far no assumption has been made for the values of real gap functions @xmath545 and @xmath546 . as will be discussed in app .
[ appendix : pfaffian ] , in order to be consistent with the phase rule [ eqs .
( [ phase1 ] ) and ( [ phase2 ] ) ] and for @xmath49 to be invariant under the transformation @xmath365 ( sec .
[ secresults : sondhi ] ) , the gap functions @xmath545 and @xmath546 are subject to the following conditions : @xmath567 { \bar{\delta}}_{{\vec t}_m } = 0 , & ~~ { \rm for~m_2~even}. \end{array } \right.\ ] ] with this condition , @xmath350 given in eq .
( [ pairbroken ] ) is equivalent to @xmath545 and @xmath546 defined here .
let us now consider one special but important case for which all hopping terms @xmath544 are zero and @xmath568 .
in this case , @xmath569 , and therefore the excitation spectrum for the bogoliubov mode is simply @xmath570^{1/2}.\ ] ] after tedious but straight forward calculations , the ground state @xmath64 of the bcs hamiltonian @xmath49 for this special case can be calculated analytically : @xmath571|0\rangle } \end{array}\ ] ] where @xmath572 and @xmath573 is used .
this state should be compared with the more conventional one given by eq .
( [ wfbcs ] ) where there is only one site per unit cell .
finally , let us consider two more specific cases separately .
_ case _ ( 1 ) : @xmath574 and only nearest neighbor gap functions are finite , _
i.e. _ , @xmath575 ( see fig .
[ lattice3 ] ) . because @xmath576 and @xmath577 $ ] , @xmath561 has gapless excitations at @xmath578 and @xmath579 .
_ case _ ( 2 ) : @xmath580 . since @xmath581 and @xmath582 in this limit , _
i.e. _ , a pairing function is proportional to a gap function , a projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] built from this bcs state becomes @xmath583{1.25mm}{0.25mm}}{\rm bcs}\rangle}={\cal p}_{\rm g}\left[\sum_{\vec r } \left\ { { \sum_{{\vec t}_m}}^{\prime } \left ( \delta_{{\vec t}_m } + ( -1)^{r_1}{\bar{\delta}}_{{\vec t}_m}\right ) \left ( c^\dag_{{\vec r}{\uparrow } } c^\dag_{{\vec r}+{\vec t}_m{\downarrow } } -c^\dag_{{\vec r}{\downarrow } } c^{\dag}_{{\vec r}+{\vec t}_m{\uparrow } } \right ) \right\ } \right]^{l/2}|0\rangle . } \ ] ] from this result , it it clear that the projected bcs state discussed here indeed includes a short range rvb wave function @xmath392 since if @xmath584 ( all other gap functions are zero ) is chosen , the above wave function ( [ pbcs - svb ] ) consists of the pairing functions with exactly the same phase as in eqs .
( [ phase1 ] ) and ( [ phase2 ] ) .
more details of this relation will be found in the next appendix .
we start from the definition of a pfaffian in terms of an antisymmetric @xmath585 matrix @xmath586 where @xmath587 .
the pfaffian of matrix @xmath588 is @xmath589=\sum\limits_{{\scriptstyle ( i_1<j_1),(i_2<j_2),\cdots , ( i_n < j_n ) } \atop { \scriptstyle~{\rm and}~ i_1
< i_2 \cdots < i_n } } ( -1)^p \prod\limits_{k=1}^n f_{i_k , j_k},\ ] ] where the sum runs over all possible covering of indices @xmath590 such that @xmath591 and @xmath592 , being @xmath139 the parity of the permutation of the @xmath593 indices : @xmath594 the most important relation known for the pfaffian is that @xmath595\right)^2={\rm det}[{\bar f}]$ ] , which however will not be used in the following . now let us suppose that the indices @xmath596 label the positions of a lattice ( not necessarily one dimensional ) .
then each covering of the indices in the pfaffian is interpreted as a particular dimer configuration in which for example a spin singlet pair located at sites @xmath597 is assigned .
where @xmath599 is the spin-@xmath600 lowering operator at site @xmath36 , @xmath601 is the ferromagnetic state defined by @xmath602 and @xmath603 is the electron vacuum .
notice that for convenience we have included in the definition of the wave function the same permutation sign @xmath604 appearing in the pfaffian ( [ pfaf ] ) .
the second wave function is the projected bcs state : @xmath605{1.25mm}{0.25mm}}{\rm bcs}\rangle}= { \cal p}_{\rm g } \exp\left [ { \sum_{\scriptstyle{i , j=1 } \atop { \scriptstyle{(i <
j ) } } } ^{2n } f_{i , j } \left ( c^{\dag}_{i,\uparrow } c^{\dag}_{j,\downarrow } - c^{\dag}_{i,\downarrow } c^{\dag}_{j,\uparrow } \right ) } \right ] |0\rangle\ ] ] where @xmath42 is the gutzwiller projection operator onto singly occupied sites . in both wave functions
, we have used the upper triangular part of the antisymmetric matrix @xmath588 , whereas the tedious permutation sign @xmath604 is present only in the wave function @xmath606 .
we shall now show that the two wave functions @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] and @xmath607 are actually the same _ for any @xmath588 and on any lattice with all possible boundary conditions_. 0.5 truecm * proof .
* because the projection @xmath42 forbids double occupancy and thus singlet bonds sharing the common site , the projected bcs wave function @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] can be expanded for all dimer coverings as follows : @xmath608{1.25mm}{0.25mm}}{\rm bcs}\rangle}= { \frac{1}{n!}}{\cal p}_{\rm g } \left [ \sum_{\scriptstyle{i , j=1 } \atop { \scriptstyle{(i < j ) } } } ^{2n } f_{i , j } ( c^{\dag}_{i,\uparrow } c^{\dag}_{j,\downarrow } - c^{\dag}_{i,\downarrow } c^{\dag}_{j,\uparrow } ) \right]^n |0\rangle = \sum\limits_{{\scriptstyle ( i_1<j_1),(i_2<j_2),\cdots , ( i_n < j_n ) } \atop { \scriptstyle~{\rm and}~ i_1 < i_2 \cdots < i_n } } \prod\limits_{k=1}^n \left [ f_{i_k , j_k } ( c^{\dag}_{i_k,\uparrow } c^{\dag}_{j_k,\downarrow } - c^{\dag}_{i_k,\downarrow } c^{\dag}_{j_k,\uparrow } ) \right ] |0\rangle,\ ] ] where in the last formula the constant @xmath609 cancels out the factor due to the ordering of the indices @xmath610 . on the other hand , by substituting the definition of the ferromagnetic state in the wave function @xmath607 and making the necessary @xmath611 fermion permutations for each dimer covering , we arrive at the following expression : by simple inspection , it is readily realized that the fermion sign @xmath613 is exactly the one @xmath604 defining the permutation sign of the covering of indices @xmath614 , and that @xmath615 .
thus it is proved that two wave functions @xmath607 and @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] are exactly the same . in the remaining of this appendix
, we will show that even for the triangular lattice geometry , the so - called short range rvb wave function @xmath392 [ eq .
( [ srrvb ] ) ] can be described by the projected bcs wave function ( [ pbcs ] ) with a particular choice of the matrix @xmath586 .
the short range rvb wave function @xmath392 [ eq .
( [ srrvb ] ) ] is defined with the same weight for all nearest neighbor valence bonds .
therefore , if we can find the matrix @xmath586 in eq .
( [ rvb ] ) in such a way that the sign of the permutation for each dimer covering is exactly canceled , _
i.e. _ , @xmath616 then a relation is established between rvb wave functions and projected bcs wave functions . the condition ( [ condition ] )
is highly non trivial and difficult to satisfy as the number of dimer coverings is exponentially large and the entries of the matrix are few in comparison .
nevertheless , this problem for the case of nearest neighbor dimer covering was solved in all planar graphs .
@xcite by applying these old results , @xcite we can generalize the read - chakraborty relation between the short range rvb and projected bcs wave functions on the square lattice @xcite for the triangular lattice case .
namely , by using the known matrix @xmath588 reported in ref . , it is possible to satisfy the condition ( [ condition ] ) even for the triangular case .
@xcite here we use open boundary conditions for a lattice of @xmath617 sites ( @xmath287 being multiple of @xmath618 ) where site @xmath36 is ordered lexicographically : @xmath619 .
the cartesian coordinates of the triangular lattice ( fig .
[ lattice ] ) are thus @xmath620 the pairing functions @xmath621 which meet the condition ( [ condition ] ) reads : @xmath622 1 & ~\mbox{for $ j=(0,1)$ } \\ [ 0.1 cm ] 1 & ~\mbox{for $ j=(-1,1)$ } \end{array } \right.\ ] ] for @xmath623 , whereas @xmath624 -1 & ~\mbox{for $ j=(1,1)$ } \\ [ 0.1 cm ] -1 & ~\mbox{for $ j=(0,1)$ } \end{array } \right.\ ] ] for @xmath625 .
all the other values of @xmath621 are obtained by translation of @xmath626 and/or @xmath627 .
therefore , the unit cell of @xmath621 is @xmath362 on the triangular lattice ( see fig . [ lattice3 ] ) .
we can multiply by the imaginary unit @xmath36 all the @xmath628 for the odd @xmath629 components of @xmath335 since the resulting wave function is equivalent to the original one [ eq .
( [ pbcs ] ) ] in the presence of the projection operator @xmath42 ( apart from an overall phase ) .
the obtained new complex matrix @xmath586 consists of : @xmath630 1 & ~\mbox{for $ j=(0,1)$ } \\ [ 0.1 cm ] i & ~\mbox{for $ j=(-1,1)$ , the diagonal bond } \end{array } \right.\ ] ] for @xmath623 , whereas @xmath631 1 & ~\mbox{for $ j=(1,1)$ } \\ [ 0.1 cm ] -i & ~\mbox{for $ j=(0,1)$ , the other diagonal bond } \end{array } \right.\ ] ] for @xmath625 .
all the other values of @xmath621 are obtained by translation of @xmath626 and/or @xmath627 .
notice that if the diagonal bonds are eliminated , we end up with again a @xmath632 unit cell with @xmath633 in fourier space , thus recovering the read - chakraborthy result for the square lattice with @xmath634 .
@xcite it is important to emphasize that for the triangular lattice geometry a @xmath89 unit cell is inevitable even with the complex representation ( as the diagonal bonds acquire different signs ) .
thus , we conclude that the translation symmetry in @xmath621 has to be broken for a good projected bcs wave function in the triangular case .
it is interesting that two sites per a unit cell is enough , in contrast to the classical nel order state with @xmath635 degrees of mutual spin orientations which contains instead three sites per unit cell . for the short range
rvb state with periodic boundary conditions , it is known that condition ( [ condition ] ) can not be satisfied by a single @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] , but by four , all obtained with the possible choices of periodic and antiperiodic boundary conditions in @xmath636 and @xmath637 directions .
@xcite the proof can be applied for each of such @xmath41{1.25mm}{0.25mm}}{\rm bcs}\rangle}$ ] , showing that these four projected bcs wave functions have to be used to exactly match the short range rvb wave function with periodic boundary conditions .
r. coldea , d. a. tennant , a. m. tsvelik , and z. tylczynski , , 1335 ( 2001 ) ; r. coldea , d. a. tennant , k. habicht , p. smeibidl , c. wolters , and z. tylczynski , , 137203 ( 2002 ) ; r. coldea , d. a. tennant , and z. tylczynski , 134424 ( 2003 ) .
when @xmath643 , the off - diagonal pairing terms in eq .
( [ bcs2x1 ] ) are completely eliminated by the following unitary transformation : @xmath644 c_{-{\vec k}\downarrow}^\dag \end{array } \right ) = { \hat{\cal u}}_y(\pi/2 ) \left ( \begin{array}{c } c_{{\vec k}\uparrow}\\[0.1 cm ] c_{-{\vec k}\downarrow}^\dag \end{array } \right).\ ] ] here @xmath645 and @xmath646 is the @xmath401-component of the pauli matrices .
this property has been verified by numerically computing both diagonal and off diagonal spin - spin correlation functions .
namely , the mixed average estimate of the magnetic order parameter in the @xmath647-plane was computed and found to remain negligibly small .
fne provides results similar to gfmcsr , @xcite when the same guiding function is used . however , fne is much more efficient , as far as the statistical errors are concerned , and provides a variational upper bound of the energy .
the way to open a finite spin excitation gap discussed here is similar to the one realized in the quantum dimer model on the triangular lattice , @xcite where in order to treat the dimer covering of the lattice , an algebraic method based on the pfaffian was introduced .
the corresponding matrix is also defined in a @xmath369 unit cell ( see also app .
[ appendix : pfaffian ] ) . | the quantum spin-1/2 antiferromagnetic heisenberg model on a two dimensional triangular lattice geometry with spatial anisotropy is relevant to describe materials like @xmath0 and organic compounds like @xmath1-(et)@xmath2cu@xmath2(cn)@xmath3 .
the strength of the spatial anisotropy can increase quantum fluctuations and can destabilize the magnetically ordered state leading to non conventional spin liquid phases . in order to understand these intriguing phenomena , quantum monte carlo methods
are used to study this model system as a function of the anisotropic strength , represented by the ratio @xmath4 between the intra - chain nearest neighbor coupling @xmath5 and the inter - chain one @xmath6 .
we have found evidence of two spin liquid regions .
the first one is stable for small values of the coupling @xmath7 , and appears gapless and fractionalized , whereas the second one is a more conventional spin liquid with a small spin gap and is energetically favored in the region @xmath8 .
we have also shown that in both spin liquid phases there is no evidence of broken translation symmetry with dimer or spin - peirls order or any broken spatial reflection symmetry of the lattice .
the various phases are in good agreement with the experimental findings , thus supporting the existence of spin liquid phases in two dimensional quantum spin-1/2 systems . |
as in the previous paper @xcite we define the system @xmath4 as an aggregation of particles capable for autocatalytic reactions .
symbols @xmath3 and @xmath1 are used for notations of _ autocatalytic and substrate particles _ , respectively .
we denote by @xmath5 _ the particles of end - product _ which do not take part in the reactions .
we assume that the system is `` open '' for particles @xmath1 , i.e. the number of particles @xmath1 is kept constant by a suitable reservoir . however , the system is strictly closed for the autocatalytic particles @xmath3 . in this paper
we investigate systems which are governed by reactions @xmath6 where @xmath7 and @xmath8 are the rate constants .
these reactions became interesting since convincing speculations were published @xcite about the `` natural selection '' based on their properties .
the organization of the paper as follows .
after brief discussion of the kinetic rate equation , in section i we derive and formally solve the forward kolmogorov equation for the probability @xmath9 of finding @xmath10 particles in a system of volume @xmath11 at time moment @xmath12 . defining the conditions of the stationarity we determine the stationary probability @xmath13 , and analyze its properties . in section
ii we present a modified stochastic model of the autocatalytic reactions ( [ a1 ] ) which is capable of reproducing the solution of the rate equation , however , brings about such a large fluctuation in the stationary number of @xmath3 particles , that the mean value loses practically all of its information content . in section
iii we define the lifetime of a system controlled by reaction ( [ a1 ] ) , and calculate exactly the extinction probability , as well as the mean value of the system lifetime .
finally , in section iv we summarize the main conclusions .
in order to make comparison , first the well - known results of the kinetic rate equation are briefly revisited , and then the stochastic model of the reactions ( [ a1 ] ) will be thoroughly analyzed .
let @xmath14 be the number of @xmath3 particles in the volume @xmath11 of the system @xmath4 at the time moment @xmath12 , and denote by @xmath15 the number - density of the @xmath3 particles and by @xmath16 that of @xmath1 particles which is kept strongly constant by suitable reservoir . according to the kinetic law of mass action
we can write @xmath17,\ ] ] where @xmath18 is the _ critical parameter _ of the reaction . taking into account the initial condition
@xmath19 we obtain immediately the solution of ( [ z1 ] ) in the form : @xmath20 clearly we have two stationary solutions , namely @xmath21 it is easy to show that the solution of @xmath22 is stable , if @xmath23 , i.e. if @xmath24 , while @xmath25 is stable , if @xmath26 , i.e. if @xmath27 .
let @xmath28 a small disturbance in @xmath29 .
it follows from eq .
( [ z1 ] ) that @xmath30\;\delta c,\ ] ] and so , we see immediately , if @xmath31 , then @xmath32 , i.e. @xmath22 is stable , when @xmath33 . similarly , if @xmath34 , then @xmath35 , i.e. @xmath25 is stable , when @xmath36 .
the stationary density of @xmath3 particles versus @xmath16 can be seen in fig .
it is clear that @xmath37 is a `` bifurcation point '' , since if @xmath38 then there are two stationary solutions but among them only one , namely the solution @xmath22 is stable . by decreasing the density @xmath16 _ adiabatically _ below the critical value @xmath39 , the autocatalytic particles @xmath3 are dying out completely , and it is impossible to start again the process by increasing the density @xmath16 above the critical @xmath39 .
( 5,6)(0,0)(0,1)5 ( 0,0)(1,0)6 ( 2,0)(3,4)3 ( 0,0)(1,0)2 ( 0,5.5)(0,0)[t]@xmath40 ( 6.5,0)(0,0)[t]@xmath16 ( 2,-0.5)(0,0)@xmath39 ( 1 , 2.5)(0,-1)2.5 ( 5 , 2.5)(-1,0)1.1 ( 1.1,3)(0,0)[t]@xmath25 ( 5.4,2.8)(0,0)[t]@xmath22 fig.@xmath41 : dependence of the stationary number - density of @xmath3 particles on the number - density @xmath16 .
the thick lines refer to stable stationary values . in the next subsection
we will analyze the stochastic model of reversible reactions ( [ a1 ] ) .
it will be shown that in the stochastic model the long time behavior of the number of autocatalytic particles is completely different from that we obtained by using the rate equation ( [ z1 ] ) .
it is to mention that interesting and seemingly convincing speculations were published @xcite about the possibility of `` natural selection '' based on autocatalytic reactions in sets of prebiotic organic molecules and about the `` origin of life '' beginning with a set of simple organic molecules capable of self - reproduction .
the essence of these speculations can be summarized as follows : let us consider @xmath42 different and independent autocatalytic particles , and denote by @xmath43 the corresponding bifurcation points . if @xmath44 , then the stationary density of all autocatalytic particles is larger than zero .
when the density @xmath16 decreases _ adiabatically _ to a value lying in the interval @xmath45 , then the autocatalytic particles @xmath46 disappear from the system , and by increasing again the density of @xmath1 particles above @xmath47 , there is no possibility to recreate those particles which were lost . in this way some form of selection can be realized .
when we take into account the stochastic nature of the autocatalytic reaction , then we will see that speculations of this kind can not be accepted .
let the random function @xmath48 be the number of autocatalytic particles @xmath3 at the time moment @xmath12 .
the task is to determine the probability @xmath49 of finding exactly @xmath50 autocatalytic particles @xmath3 in the system @xmath4 at time instant @xmath12 provided that at @xmath51 the number of @xmath52 particles was @xmath53 .
assume that the number of the substrate particles @xmath1 is kept constant during the whole process , we can write for @xmath54 the equation : @xmath55 + \ ] ] @xmath56p_{n+1}(t)\ ; \delta t + o(\delta t),\ ] ] and for @xmath57 the equation : @xmath58 where @xmath59 from these equations it follows immediately that @xmath60 @xmath61 where @xmath62 .
if @xmath57 , then @xmath63 for the sake of completeness we derive the generating function equations @xmath64 and @xmath65 by using the equations ( [ z7 ] ) and ( [ z8 ] ) it is easy to show that @xmath66 satisfies the partial differential equation @xmath67 while @xmath68 the equation @xmath69 \ ; \frac{\partial g_{exp}(t , y)}{\partial y } - \beta ( 1-e^{-y})\ ; \frac{\partial^{2 } g_{exp}(t , y)}{\partial y^{2}}.\ ] ] the initial conditions are @xmath70 and @xmath71 , respectively , and in addition it is to note that @xmath72 . for many purposes
it is convenient to use the logarithm of the exponential generating function @xmath73 ) .
therefore , define the function @xmath74 the derivatives of which at @xmath75 , i.e. @xmath76_{y=0 } = \kappa_{j}(t ) , \;\;\;\;\;\ ; j = 1 , 2 , \ldots\ ] ] are the _ cumulants _ of @xmath48 .
one can immediately obtain the equation @xmath77\;\frac{\partial k(t , y)}{\partial y } - \beta ( 1-e^{-y})\ ; \left\{\frac{\partial^{2 } k(t , y)}{\partial y^{2 } } + \left[\frac{\partial k(t , y)}{\partial y}\right]^{2}\right\}.\ ] ] now , the initial condition is @xmath78 in order to simplify the notations define the vector @xmath79 and the matrix @xmath80 where @xmath81 and write the eqs .
( [ z7 ] ) and ( [ z8 ] ) in the following concise form : @xmath82 the formal solution of this equation is @xmath83 where the components of @xmath84 are @xmath85 and the matrix @xmath86 is a _
normal jacobi matrix_. as known , the eigenvalues of a normal jacobi matrix are different and real . if @xmath87 are the eigenvalues of @xmath86 , then the @xmath50th component of the vector @xmath88 can be written in the form @xmath89 for any @xmath90 .
since to find the eigenvalues @xmath91 and the coefficients @xmath92 is not a simple task , we concentrate our efforts only on the determination of stationary solutions of eqs .
( [ z7 ] ) and ( [ z8 ] ) .
in order to show that the limit relations @xmath93 exist , we need a _
theorem for eigenvalues of the matrix @xmath86 _ stating that @xmath94 and @xmath95 for every @xmath96 .
the proof of the theorem can be found in * appendix a*. by using this theorem it follows from eq .
( [ z20 ] ) that @xmath97 since in this case @xmath98 we can write from ( [ z7 ] ) immediately the stationary equations @xmath99 = n [ ( n+1 ) w_{n+1 } - ( n-1)w_{n } ] + b\;[(n+1 ) w_{n+1 } - n w_{n}],\ ] ] for all @xmath100 . summing up both sides of ( [ z23 ] ) from @xmath101 to @xmath102 , we obtain @xmath103 taking into account the eq .
( [ z8 ] ) we should write that @xmath104 = w_{0 } + \sum_{k=1}^{\infty } w_{0k}\ ; e^{-|\nu_{k}|u},\ ] ] and we see that the condition of the stationarity is either @xmath105 , or @xmath106 , i.e. @xmath107 . if @xmath108 , then it follows from ( [ z24 ] ) that @xmath109 for all @xmath110 , and consequently , @xmath111 since @xmath112 if @xmath113 , i.e. if the autocatalytic particles @xmath3 do not decay , then @xmath114 and from ( [ z24 ] ) one obtains that @xmath115 taking into account that @xmath116 we can write @xmath117 for @xmath54 , while for @xmath57 we have @xmath118 for the calculation of factorial moments let us introduce the generating function @xmath119 in order to obtain the equation for the mean value of the number of @xmath3 particles we use the generating function equation ( [ z10 ] ) . introducing the time parameter @xmath120 we have @xmath121 - \beta\;[m_{2}(u)-m_{1}^{2}(u)],\ ] ] which clearly shows that the kinetic law of the mass action is violated , when the variance @xmath122 is not negligible . for the second moment
we get the equation @xmath123\;m_{2}(u ) - 2\;m_{3}(u),\ ] ] in which appears the third moment @xmath124 .
there are several methods to find approximate solution of @xmath125 , some of them were mentioned already in @xcite . here
, we do not want to discuss the details , instead we are focussing our attention on the properties of _ the expectation value and variance in stationary state_. if the decay constant @xmath126 of @xmath3 particles is zero , then it is easy to show from eq .
( [ z28 ] ) that the stationary value of the average number of autocatalytic particles in the system is equal to @xmath127 while _ the second factorial moment _ is @xmath128 consequently
, the variance can be written in the form @xmath129 = m_{1}^{(st)}\left(1 - \frac{a}{e^{a } - 1}\right).\ ] ] if @xmath130 , then @xmath131 .
the first important conclusion drawn from the stochastic model is that the average number of the autocatalytic particles in stationary state is different from zero , only when the decay rate constant @xmath126 of @xmath3 particles is zero .
it means that _ there is no `` bifurcation '' point in the dependence of the average stationary number of @xmath3 particles on the number of @xmath1 particles _ , therefore speculations mentioned earlier about the `` natural selection '' are not supported by the stochastic model .
the second conclusion is connected with _ the law of the mass action _ referring to the _ chemical equilibrium_. if @xmath132 , then the reversible reaction @xmath133 has to lead to an equilibrium state , in which @xmath134 the stochastic model results in an entirely different expression , namely @xmath135 the relative dispersion , i.e. @xmath136 clearly shows that the fluctuations of the number of autocatalytic particles become poisson - like when the number of substrate particles is increasing .
the modification of the stochastic model is very simple .
we assume that the probability of the reverse reaction @xmath137 is proportional to the average number of @xmath3 particles .
therefore , the probability that a reverse reaction occurs in the time interval @xmath138 is nothing else than @xmath139 provided that the number of @xmath3 particles was exactly @xmath50 at time moment @xmath140 . accepting this assumption
we can rewrite instead of @xmath9 . ] the equations ( [ z7 ] ) and ( [ z8 ] ) in the form : @xmath141\ ; n\;\tilde{p}_{n}(t ) + \alpha n_a(n-1)\;\tilde{p}_{n-1}(t ) + \ ] ] @xmath142(n+1 ) \;\tilde{p}_{n+1}(t ) , \;\;\;\;\;\ ; n = 1 , 2 , \ldots,\ ] ] and @xmath143 respectively .
one can immediately see that the generating function @xmath144 satisfies the equation @xmath145(1 - e^{-y})\right\}\;\frac{\partial \tilde{g}_{exp}(t , y)}{\partial y},\ ] ] with @xmath146 and @xmath147 . in the sequel @xmath148 . introducing the notations @xmath149 and @xmath120 , it is not surprising that the first moment @xmath150 is the solution of the equation @xmath151,\ ] ] which is exactly the same as ( [ z1 ] ) .
if the initial condition is @xmath152 , then @xmath153 and one can see that @xmath154 if @xmath155 , i.e. if @xmath156 , then one obtains from ( [ d5 ] ) @xmath157 the value @xmath158 is the number of @xmath1 particles that corresponds to _ the bifurcation concentration @xmath39 _ introduced in the rate equation model .
however , the modified stochastic model takes into account the randomness of the reactions , and hence gives a possibility for the determination of _ the variance of the number of @xmath3 particles _ versus time .
it can be easily shown that the variance of @xmath159 is @xmath160.\ ] ] in order to prove this relation we need the equation for the second moment @xmath161 , which can be derived from eq .
( [ d3 ] ) . introducing the time parameter @xmath120
we obtain @xmath162\;m_{1}(u ) + 2[a - b - m_{1}(u)]\;m_{2}(u),\ ] ] and this can be rewritten in the form @xmath163 it is an elementary task to show that @xmath164 and from this we obtain immediately the variance ( [ d7 ] ) . taking into account the expression ( [ d5 ] ) for @xmath150 the variance of the number of @xmath3 particles
can be written in the following form @xmath165 , & \text{if $ n_a \neq n_a^{(bf)}$ , } \\
\text { } & \text { } \\ \left[b u^{2 } + ( 2b + 1 ) u \right]/(1 + u)^{2 } , & \text{if $ n_a = n_a^{(bf)}$ , } \end{array } \right.\ ] ] where @xmath166.\ ] ] from ( [ d7 ] ) we can conclude that @xmath167 [ ht ! ] 0.2 cm [ ht ! ] 0.2 cm in order to have some insight into the nature of the time behavior of the variance of the number of autocatalytic particles we have calculated the variance @xmath168 versus time curves for different values of the number of @xmath1 particles and for several decay rate constants @xmath126 .
[ fig1 ] shows the time dependence of @xmath168 when @xmath169 at a fixed value of rate constant .
we see that the variance versus time curves have a maximum and after that they decrease slowly to zero . in contrary to this , when @xmath170 , then the variance of the number of autocatalytic particles increase monotonously to infinity .
it can be easily shown that @xmath171 and so , we can use for large @xmath172 the asymptotic formula @xmath173 the time dependence of the variance @xmath174 in the case when @xmath170 is seen in fig .
[ fig2 ] for three decay rate constants .
the variance converges to the infinity linearly with increasing time parameter , if @xmath170 , and to a constant value @xmath175 , if @xmath176 , so we can say that the fluctuation of the number of @xmath3 particles in the stationary state near the `` bifurcation '' point alters the possible conclusions based only on the average number of autocatalytic particles .
we say that a system is in the state @xmath177 , when @xmath178 .
obviously , the system is live at time instant @xmath179 , when @xmath180 .
let us define the probability of transition @xmath181 by @xmath182 since the process is homogenous in time @xmath183 , and using eq .
( [ z7 ] ) we can write that @xmath184 where @xmath185.\ ] ] we see that the first equation is @xmath186 and the initial condition is given by @xmath187 .
the random time @xmath188 due to the transition @xmath189 is _ the lifetime of the system _ which has been at @xmath190 in the state @xmath191 . it is evident that @xmath192 where @xmath193 is the probability that the lifetime @xmath188 is not larger than @xmath140 .
the moments of the lifetime are given by @xmath194 the extinction of a system of state @xmath191 occurs when a transition to the state @xmath195 is realized for any @xmath12 .
we define the extinction probability by @xmath196 in accordance with ( [ z27 ] ) we find that @xmath197 it is a remarkable result stating that a system being in any of states @xmath198 at a given time instant @xmath199 after elapsing sufficiently long time @xmath200 will be almost surly annihilated , if @xmath130 , and never dies , i.e. the extinction probability is zero , if @xmath132 .
it is to mention that the statement ( [ z40 ] ) can be obtained from a nice lemma by karlin @xcite which can be formulated in the following way : introducing the notation @xmath201 where @xmath202 and @xmath203 are non - negative real numbers which are not necessarily equal to ( [ z37 ] ) , it can be stated that if @xmath204 while if @xmath205 by using the expressions ( [ z37 ] ) we see immediately that @xmath206 hence we can conclude that @xmath207 and so , applying the lemma we prove the statement ( [ z40 ] ) . to determine the transition probability @xmath208 , i.e. the probability @xmath193 is not an easy problem . instead
, we show how to calculate the average lifetime @xmath209 .
let us define the parameters @xmath210 and formulate the following statement called karlin s theorem @xcite .
_ if @xmath211 , then the average lifetime @xmath212 of a system containing @xmath50 autocatalytic particles is given by @xmath213 where @xmath214 is defined by ( [ z41 ] ) , and in contrary , if @xmath215 , then @xmath216 _ the proof of this statement can be found in * appendix b*. now , by using this statement we would like to calculate the average lifetime of a system which is in the state @xmath217 at the moment @xmath140 . introducing the notations @xmath218 and by using the expressions ( [ z37 ] ) we can write @xmath219 where @xmath220 is the kronecker - symbol .
define the sum @xmath221 for @xmath100 .
if @xmath130 , then one can see immediately that @xmath222 i.e. the formula ( [ z45 ] ) should be used for the calculation of the average lifetime @xmath212 .
first , determine @xmath223 .
it follows from eq .
( [ z45 ] ) that @xmath224 where @xmath225 is the confluent hypergeometric function .
the next step is the calculation of the expression @xmath226 which can be rewritten into the form : @xmath227 after some elementary algebra we obtain @xmath228 and finally we have @xmath229 introducing a new index @xmath230 we have @xmath231 which can be transformed into the expression @xmath232 by using the identity @xmath233 @xmath234 takes a new form , namely @xmath235\;dv.\ ] ] taking into account this formula the expression ( [ z50 ] ) can be rewritten in the form @xmath236\;dv,\ ] ] which is convenient for numerical calculations . from this equation
we see that @xmath237 and we can prove the limit relation @xmath238 [ ht ! ] 0.2 cm it is _ necessary and sufficient _ to show that @xmath239 since @xmath240 is a non - negative , monotonously increasing function of @xmath241 we can write immediately that @xmath242 and if @xmath243 , then @xmath244 consequently , we obtain the inequality @xmath245 and this proves the statement ( [ z52 ] ) .
we calculated how the mean value @xmath223 depends on the ratio @xmath246 at three different values of the number of @xmath1 particles .
the results are shown in fig .
we can see that @xmath247 decreases rapidly with increasing @xmath246 , and if the values @xmath246 are smaller than @xmath248 , then we can observe that the larger is the number @xmath249 in the system the longer is the average lifetime @xmath247 .
[ ht ! ] 0.2 cm that fig .
[ fig5 ] shows is rather surprising .
the mean value of the system lifetime does not depend practically on the number of @xmath3 particles to be found in the system at that time moment which the lifetime is counted from .
we can state that the average lifetime of systems controlled by reactions ( [ a1 ] ) is already determined by several @xmath52 particles , and even a large increase of the number of @xmath52 particles does not effect significantly on the system lifetime .
[ ht ! ] 0.2 cm on this basis imagine an `` organism '' which consists of @xmath52 particles capable of self - reproduction and self - annihilation . assume that the organism becomes dead if it loses the last @xmath52 particle .
one might think that the greater is the number of @xmath52 particles in the organism the larger is irs average lifetime .
contrary to this conviction , an organism containing , let us say @xmath250 particles hardly lives longer than that which contains only @xmath251 particles . by using the values @xmath252 and @xmath253 we obtain that @xmath254 and @xmath255 .
it is rather surprising that the increase is only @xmath256 . in fig .
[ fig6 ] we see the dependence of the logarithm of the mean value @xmath257 on the number of @xmath1 particles at three different @xmath246 .
one can observe that the increase of @xmath249 results in a rather large lengthening of the average lifetime , i.e. the effect of the substrate particles on the process is much stronger than that of the autocatalytic particles .
we assumed the distribution of reacting particles in the system volume to be uniform and introduced the notion of _ the point model of reaction kinetics_. in this model the probability of a reaction between two particles per unit time is evidently proportional to the product of their actual numbers . by using this assumption
we constructed a stochastic model for systems controlled by the reactions @xmath258 provided that the number of @xmath1 particles is kept constant by a suitable reservoir , and the end - product particles @xmath5 do not take part in the reaction .
we have shown that the stochastic model results in an equation for the expectation value @xmath125 of autocatalytic particles @xmath3 which differs strongly from the kinetic rate equation .
further , we found that if the decay constant @xmath126 of the particles @xmath3 is not zero , then _ the stochastic description _ , in contrary to the rate equation description , _ brings about only one stationary state with probability @xmath259 _ , and it is the zero - state @xmath260 .
it has been also proven that the probability of a nonzero stationary state is larger than zero , if and only if the decay rate constant is equal to zero .
consequently , the average number of @xmath3 particles in the stationary state is larger than zero , if only @xmath132 .
however , one has to underline that this average number is completely different from that which corresponds to the law of the mass action of reversible chemical reactions .
we paid a special attention on the random behavior of _ the system lifetime _ , and derived an exact formula for the average lifetime .
it has been shown that the mean value of the system lifetime does not depend practically on the number of @xmath3 particles to be found in the system at the time instant which the lifetime is counted from . for example , the lifetime of a system having @xmath250 @xmath3 particles at the beginning is larger only by @xmath261% than the lifetime of a system containing @xmath251 @xmath3 particles at the time moment @xmath190 .
as mentioned already , the eigenvalues of a normal jacobi matrix are real and different .
we would like to prove the following theorem : * theorem . * _ the @xmath262 is an eigenvalue of the normal jacobi matrix @xmath263 defined by ( [ z14 ] ) and the other nonzero eigenvalues are all negative . _
* proof . *
denote by @xmath264 is the set of nonnegative integers . ]
the elements of matrix @xmath86 .
we see immediately that @xmath265 i.e. the sum of elements of any column of the matrix @xmath86 is equal to zero . strictly speaking
, the theorem itself is a straightforward consequence of this property . if @xmath266 an eigenvalue , then @xmath267 and @xmath268 define a nonzero eigenvector @xmath269 let @xmath270 be an arbitrary nonzero vector and form the following expression : @xmath271 which can be rewritten in the form @xmath272 this equality can be valid only if @xmath273 and because @xmath270 is an arbitrary nonzero vector one can chose its components to be equal to unity . then taking into account the property ( [ c1 ] ) one has @xmath274
i.e. @xmath275 is indeed an eigenvalue .
now , we show if @xmath276 , then @xmath277 . since @xmath278 , and @xmath279 ,
if @xmath280 , it follows from ( [ c2 ] ) that @xmath281 as @xmath270 is an arbitrary nonzero vector , let us chose it so that @xmath282 we obtain the inequality @xmath283 which can be valid only if @xmath277 .
* theorem . * _ if @xmath211 , then the average lifetime @xmath212 of a system containing @xmath50 autocatalytic particles is given by @xmath284 where @xmath214 is defined by ( [ z41 ] ) , and in contrary , if @xmath215 , then @xmath216 _ * proof . * assume the system to be in the state @xmath285 at a given time instant @xmath140 and suppose that the first reaction after @xmath140 occurs at a random time moment @xmath286 .
this reaction can result in a transition to either the state @xmath287 or @xmath288 with probabilities @xmath289 respectively .
since @xmath140 is arbitrary , the equation @xmath290 is valid with probability @xmath259 . taking into account that @xmath291 one obtains from ( [ e1 ] ) the recursion @xmath292 where @xmath293 . introducing the difference @xmath294 after simple rearrangements we obtain @xmath295 the solution of which
can be written in the form : @xmath296 by using the notation @xmath297 and taking into account the identity @xmath298 and the relation @xmath299 , we have @xmath300 if @xmath301 , then @xmath302 i.e. the average lifetime of the system is infinite . the proof is simple : it is obvious that @xmath303 , therefore , it follows from ( [ e4 ] ) that @xmath304 for any @xmath50 , and if @xmath305 , than @xmath223 must be infinite . since @xmath306 , it is evident that @xmath307 , hence the statement is proven . if @xmath308 , then one can find a finite real number @xmath309 such that @xmath310 , consequently @xmath311 and it follows from this that @xmath312 taking into account this relation we can rewrite eq .
( [ e4 ] ) into the form : @xmath313 where @xmath314 is defined by ( [ z41 ] ) . introducing the notation @xmath315 we have @xmath316 the solution of which
is nothing else than @xmath317 and by substituting @xmath223 and @xmath318 we obtain immediately the equation ( [ e0 ] ) . | we analyzed the stochastic behavior of systems controlled by autocatalytic reaction @xmath0 provided that the distribution of reacting particles in the system volume is uniform , i.e. the point model of reaction kinetics introduced in arxiv : cond - mat/0404402 can be applied . assuming the number of substrate particles @xmath1 to be kept constant by a suitable reservoir
, we derived the forward kolmogorov equation for the probability of finding @xmath2 autocatalytic particles @xmath3 in the system at a given time moment .
we have shown that the stochastic model results in an equation for the expectation value of autocatalytic particles @xmath3 which differs strongly from the kinetic rate equation .
it has been found that not only the law of the mass action is violated but also the bifurcation point is disappeared in the well - known diagram of @xmath3 particle- vs. @xmath1 particle - concentration .
therefore , speculations about the role of autocatalytic reactions in processes of the `` natural selection '' can be hardly supported . |
close binary stellar systems can exchange mass , when at least one of the stars fills its roche lobe . after a critical mass @xmath6 of unburned transferred matter
is accumulated on the surface of the accreting star , ignition sets in , typically under degenerate conditions when the accreting object is a white dwarf or neutron star .
degenerate conditions for which the pressure is not a function of temperature , prevent temperature adjustment via pressure increase and expansion and cause a thermonuclear runaway and explosive burning .
for white dwarfs , a layer of @xmath7@xmath8 m@xmath9 forms , before pycnonuclear ignition of hydrogen burning sets in ( e.g. ; ; ) and causes a nova event ( explosive h - burning on white dwarfs ) with maximum temperatures of @xmath10(23)@xmath11 k and a total energy release of @xmath12@xmath13 ergs .
the burning takes 1001000 s before the partially burned hydrogen envelope is ejected .
x - ray bursts ( for an observational overview see 1993 ) involve accreting neutron stars with an unburned , hydrogen - rich surface layer .
the critical size of the hydrogen layer before ignition is as small as @xmath14 m@xmath9 .
temperatures of the order ( 1 - 2)@xmath15 k and densities @xmath16@xmath10@xmath17@xmath18 g @xmath19 are attained ( see e.g. ; 1982 ; 1983 ; 1986 ; 1993 ; 1996 ) . this explosive burning with rise times of about 110 s leads to the release of @xmath20@xmath21 ergs .
many of the observed features and general characteristics are understood , however , there is still a lack of a quantitative understanding of the detailed observational data .
another aspect is that the explosion energies are smaller than the gravitational binding energy of the accreted hydrogen envelope .
an evenly distributed explosion energy would not unbind and eject matter .
it remains to be seen whether an interesting amount of matter can escape the neutron star .
super - eddington x - ray bursts ( ) are the best candidates for this behavior . a description of hot ( explosive ) hydrogen burning has been given by ( 1980 ) , ( 1981 ) , ( 1982 ) , ( 1983 ) , ( 1986 ) , ( 1994 ) , ( 1994 ) [ see also ( 1991 ) , ( 1994 ) , ( 1992 ) , and ( 1993 ) for thorne - zytkow objects , but see the recent results by ( 1996 ) which puts doubts on their existence ] .
the burning is described by proton captures , @xmath22-decays , and possibly alpha - induced reactions on unstable proton - rich nuclei , usually referred to as the rp - process ( rapid proton capture process ) .
cross sections can either be obtained from the best available application of present experimental knowledge , e.g. the determination of resonance properties from mirror nuclei and transfer and/or charge exchange reaction studies , or actual cross section measurements , like for @xmath23n(@xmath24o , the first reaction cross section analysed with a radioactive ion beam facility ( see 1992 and references therein ) .
there exist two major motivations in nuclear astrophysics ( a ) , to understand the reaction flow to a necessary degree , in order to predict the correct energy generation required for hydrodynamic , astrophysical studies , and ( b ) , to predict a detailed isotopic composition which helps to understand the contribution of the process in question to nucleosynthesis in general .
our main motivation in this paper is ( a ) , to provide a fast energy generation network as a tool for nova and x - ray burst studies .
this might be underlined by the fact that novae and x - ray bursts seem not to be major contributors to nucleosynthesis due to the small ejected masses involved or the question whether gravitational binding can be overcome at all .
[ however , novae can be important for nuclei like @xmath25li , @xmath26n , @xmath27na , @xmath28al , and even si and s ; and super - eddington x - ray bursts ( 1996 ) will be able to eject some matter , probably containing some light p - process elements . ]
it will turn out at the end that our efficient approximation and energy generation scheme can also be used to predict abundances accurately .
the nucleosynthesis for nuclei above kr will , however , be discussed in a second paper ( et al .
1996 ) . in order to understand the energy generation correctly
, we have to be able to understand the main reaction fluxes . in the past we performed a series of rp - process studies ( 1986 to ar , and 1994 , extending up to kr ) .
the reaction rate predictions were based on resonance and direct capture contributions for proton - rich nuclei , making use of the most recent experimental data .
several ( @xmath29)-reaction rates below mass a=44 have been recalculated by ( 1995 ) in the framework of a shell model description for the level structure of the compound nucleus .
a preliminary analysis of the major aspects was given by ( 1994 ) .
in the following section [ reacflow ] we will present this in more detail and discuss the constraints which an approximation scheme has to fulfill in the whole range of temperatures occurring in explosive hydrogen burning environments , i.e. @xmath0@xmath1 k. the approximation scheme will be presented in section [ approxscheme ] , its application and comparison with full network calculations in section [ results ] , followed by concluding remarks in section [ conc ] .
at low temperatures , the rp - process is dominated by cycles of two successive proton captures , starting out from an even - even nucleus , a @xmath22-decay , a further proton capture ( into an even-@xmath30 nucleus ) , another @xmath22-decay , and a final @xmath31-reaction close to stability ( similar to the hot cno ) .
the final closing of the cycle via a @xmath31-reaction occurs because proton capture q - values increase with increasing neutron number @xmath32 in an isotopic chain for a given @xmath30 , while @xmath33-capture q - values or -thresholds show a very weak dependence on @xmath32 .
thus , a cycle closure occurs at the most proton - rich compound nucleus , which has a lower alpha than proton - threshold .
these are the even-@xmath30 and especially well bound `` alpha''-nuclei with isospin t@xmath34=@xmath35=0 , like @xmath36o , @xmath37ne , @xmath38 mg , @xmath39si , @xmath40s , @xmath41ar , @xmath42ca .
therefore , @xmath31-reactions can operate on odd-@xmath30 targets with t@xmath34=+1/2 : @xmath26n , @xmath43f , @xmath44na , @xmath45al , @xmath46p , @xmath47cl , @xmath48k .
the hydrogen burning cycles are connected at these nuclei , due to a possible competition between the @xmath2- and @xmath31-reaction .
an exception to the rule is the ne isotopic chain , where the transition happens already for the compound nucleus @xmath43ne .
thus , @xmath49f(@xmath50o is a possible reaction , bypassing the of(ne)-cycle .
therefore , only the extended cn(o)- , the nena- , mgal- , sip- , scl - cycles etc .
this is displayed in figure [ fig cycle43 ] .
the progress of burning towards heavier nuclei depends on the leakage ratio @xmath51 into the next cycle , which makes a good experimental determination of these reactions important and is possible as we deal here with stable targets . on the other hand ,
the excitation energies in the corresponding compound nuclei are of the order 8.512 mev and make statistical model approaches also reliable ( see e.g. the experiments by 1993 , 1994 and 1995 for @xmath46p(@xmath52 , and @xmath47cl(@xmath52 and the later discussion ) . increasing temperatures
allow to overcome coulomb barriers and extend the cycles to more proton - rich nuclei , which permit additional leakage via proton captures , competing with long @xmath53-decays .
the cycles open first at the last @xmath53-decay before the @xmath31-reaction via a competing proton capture , i.e. at the even-@xmath30 t@xmath34=1/2 nuclei like @xmath44 mg , @xmath45si , @xmath46s , @xmath47ar , and @xmath48ca , excluding @xmath26o ( because @xmath36f is particle unstable ) and @xmath43ne , which is bypassed by the stronger @xmath49f(@xmath54-rate .
these nuclei are only one unit away from stability and have small decay q - values and long half - lives .
the situation is illustrated in figure [ fig qval1 ] , which shows the possible break - out points and the q - values of the proton capture reactions . since the reaction proceeds towards the proton drip - line and into an odd - odd nucleus , the capture q - values of these @xmath2-reactions are small ( generally less than 2 mev ) .
this implies a too small density of excited states for employing statistical model cross sections ( see discussion in the following paragraphs ) . at about 3@xmath5510@xmath56
k essentially all such cycles are open and a complete rp - pattern of proton captures and @xmath53-decays is established , with the exception of the extended cno - cycle .
the process time is determined by the sum of @xmath53-decay and proton capture time scales , @xmath57 and @xmath58 , along the rp - path , which does generally not extend more proton - rich than t@xmath34=1 for even-@xmath30 nuclei .
the reaction time scales are dominated by the t@xmath34=1/2 even-@xmath30 nuclei close to stability , listed above . at 4@xmath5510@xmath56 k
the cno - cycle breaks open via @xmath26o(@xmath59ne .
@xmath60o(@xmath61f can also lead to a successful break - out when followed by a proton capture and @xmath5-reaction on @xmath49ne . due to higher energies ,
coulomb barriers can be more easily overcome by heavier projectiles ( like alpha captures on cno - targets ) , and proton captures can occur further away from stability .
however , in all cases the decay of t@xmath34=1 even-@xmath30 nuclei ( and at high densities the competition with @xmath5-reactions ) is a prominent part of the reaction path .
alpha - reactions are not important for @xmath30@xmath6220 , due to increasing coulomb barriers and decreasing q - values . for heavier nuclei beyond ca
, the reaction pattern seems complicated .
however , temperatures approaching @xmath63 k cause strong photodisintegrations for proton capture q - values smaller than 25@xmath64@xmath103 mev , which is comparable to q - values found for even-@xmath30 t@xmath34=1 nuclei .
[ 25 - 30@xmath64@xmath65@xmath66 is a rough criterion to find out whether photodisintegrations can counterbalance capture reactions and lead to an equilibrium situation . ]
thus , reactions breaking out of the cycles discussed above ( up to the proton drip line ) come into a @xmath2@xmath3-equilibrium along isotonic lines with equal @xmath32 . under such circumstances
the nuclear connection boils down to the necessary knowledge of nuclear masses and @xmath53-decay half - lives , similar to the r - process experiencing an @xmath67@xmath68-equilibrium in isotopic chains ( see ; ; ) .
the general situation is explained in figure [ fig netzw ] , which contains all possible reactions from the original cycles over proton captures breaking out of the cycles , the accompanying beta decays and @xmath69-reactions bridging waiting points with long decay half - lives . by comparing with figure [ fig qval1 ] , which also lists the q - values involved , figures [ fig rp_fig4a ] and [ fig rp_fig4b ]
give an indication for the minimum temperatures required to permit statistical model calculations for proton and alpha - induced reactions .
the reason is that a high density of compound nuclear states is necessary for statistical model calculations to become reliable .
the energy of the gamow peak , where the integrand in a thermonuclear cross section integral ( cross section times boltzmann distribution ) has a maximum , increases with increasing temperature .
because higher level densities are encountered at higher excitation energies corresponding to higher temperatures of the boltzmann distribution , a critical temperature exists beyond which such calculations are reliable ( for more details see 1996 ) .
we can analyse the expected behavior : reactions @xmath70 , @xmath71 , and @xmath72 of figure [ fig netzw ] have reaction q - values less than @xmath103 mev and for conditions close to 10@xmath73 k we have 25@xmath64@xmath10@xmath66 which permits a ( @xmath52@xmath3-equilibrium .
thus , although the cross sections are highly uncertain and statistical model approaches are not valid ( see figure [ fig rp_fig4a ] ) , only the q - values , i.e. masses , are needed for these nuclei . reactions @xmath74 ,
@xmath75 , @xmath76 , and @xmath77 [ ( @xmath52 and ( @xmath54 reactions ] have q - values in excess of 5 mev and in most cases high enough level densities to ensure the usage of statistical model cross sections for the appropriate temperature regime .
reaction @xmath78 , an @xmath5-reaction , provides also a sufficiently high level density to apply statistical model cross sections ( see figure [ fig rp_fig4b ] , for @xmath49ne(@xmath79 ) see ) .
reaction @xmath80 is crucial for the break - out , a statistical model approach is clearly not permitted , and a ( @xmath52@xmath3-equilibrium is not valid at the lower appropriate temperatures of about 3 - 4@xmath11 k. thus , all reactions of type @xmath80 ask strongly for experimental determination , possibly with radioactive ion beams . reactions @xmath81 , @xmath82 and all other connecting @xmath53-decays are required , preferably from experiments , otherwise from theoretical qrpa predictions .
[ fig rp_fig4a ] beyond ca and ti , this scheme simplifies into one dominated by proton captures and beta - decays , with the proton captures in a ( @xmath52@xmath3-equilibrium .
the proton drip line develops a zig - zag shape and can separate two even-@xmath30 proton - stable nuclei on the same isotone by an unstable odd-@xmath30 nucleus .
( 1995 ) have shown that such gaps can be bridged by two - proton capture reactions , proceeding through the unstable nucleus , similar to the @xmath83-reaction passing through @xmath56be .
this can extend the @xmath2@xmath3-equilibrium beyond the drip line to proton - stable `` peninsulas '' , thus avoiding long beta - decay half - lives .
while alpha - induced reactions are not important for nuclei beyond ca and ti , small q - values for @xmath84 can permit @xmath85-photodisintegrations and cause a back flow , decreasing the build - up of heavy nuclei .
after having analized the reaction patterns in the previous discussion , we have now to implement these findings into an approximation which can handle and describe all of the patterns .
considering the hydrogen and helium abundances as slowly varying , and assuming a kind of local steady flow by requiring that all reactions entering and leaving a nucleus add up to a zero flux , we will devise a scheme which can adapt itself to the whole range of conditions of interest in hot hydrogen burning .
it can automatically cover situations at low temperatures , characterized by a steady flow through all connected hot cno - type cycles and branchings feeding the next cycle , to high temperature regimes where a @xmath2@xmath3-equilibrium is established , while @xmath53-decays or @xmath5-reactions feed the population of the next isotonic line .
the details are discussed in the following section .
the thermonuclear reaction rates used in our calculations have been discussed in detail by van wormer et al .
@xcite ( _ cf . _ the appendix therein).fchu / astro.html/. ] the temporal change of the isotopic abundances @xmath86=@xmath87 can be calculated by all depleting and producing reactions
@xmath88 the first term in ( [ eq netpa1 ] ) includes all one - particle reactions ( decays or photodisintegrations ) of nuclei @xmath70 producing ( @xmath82@xmath89@xmath70 ) or destroying ( @xmath82@xmath90@xmath70 ) nucleus @xmath82 .
the @xmath91 are either decay constants or effective temperature dependent decay constants in the case of photodisintegrations .
the second sum represents all two - particle reactions between nuclei @xmath92 with @xmath93=@xmath94 .
similarly , the third term represents three particle processes with @xmath93 = @xmath95@xmath96 ( 1967 ) .
the coefficients @xmath75 include the statistical factors for avoiding double counting of reactions of identical particles .
van wormer et al .
@xcite investigated numerically which reactions must be considered in the rp - process .
apart from the temperature insensitive @xmath4-decays , the only one - particle reactions which must be especially considered at high temperatures are photodisintegrations .
the two - body reactions dominating the rp - process are @xmath2- , @xmath31- and , @xmath5-reactions .
three body reactions can be neglected in explosive hydrogen burning with the exception of the @xmath83-reaction which plays a central role in this context .
taking into account individual reactions , equation ( [ eq netpa1 ] ) can be rewritten as @xmath97 if the proton and helium abundances do not change greatly during relevant rp - process time scales , we are able to describe proton captures and ( @xmath98)-reactions with _ quasi - decay constants _ , provided that temperature and density do not change greatly during the process time scale of interest : @xmath99 thus the system of nonlinear differential equations can be written in a linearized form @xmath100 where @xmath101 represents one of the following reactions : @xmath3 , @xmath2 , @xmath5 , @xmath31 , @xmath102 , @xmath85 , @xmath103 . when we choose @xmath104=@xmath105 @xmath106 @xmath107 to represent the initial abundance distribution , the system of equations ( [ eq netpa4 ] )
can be written as @xmath108 such systems of linear differential equations with _ constant _ coefficients can be solved by jordan - normalizing matrix @xmath109 .
its transformation @xmath110 yields the @xmath32-dimensional set of eigenfunctions @xmath111 here , @xmath112 represents eigenvalue @xmath82 with degeneracy @xmath113 , @xmath114 eigenvector @xmath70 to eigenvalue @xmath82 ( @xmath115@xmath116@xmath117@xmath113 ) and @xmath118 a polynomial of degree ( @xmath70@xmath115 ) .
however , the transformations @xmath119=@xmath120 are in general difficult to survey and to handle .
furthermore , the elements of @xmath109 become slowly varying functions when the abundances of hydrogen and helium change slightly over a certain process time scale .
this is for instance the case , when we consider hot hydrogen burning .
only numeric simulations will be feasible for these two reasons .
we will show in the following subsection how certain sequences of a thermonuclear reaction network can be regarded as _ quasi - nuclei_. in such sequences a steady flow materializes and the ratios between the different abundances are constant and are only determined by the values of the quasi - decay constants .
a procedure of this kind provides us with a system of equations with a smaller dimension , which can be solved numerically . in section [ reacflow ]
, we discussed the burning cycles with the same structure as the hot cno - cycle ( see also figures [ fig cycle43][fig netzw ] ) .
it is well understood how a steady state equilibrium is established in hot cno - like reaction chains at temperatures @xmath121 and on time scales governed by the longest @xmath4-decay in the cycle .
then the abundance @xmath122 of nucleus ( @xmath82 ) in a cycle ( @xmath70 ) can be calculated from the equilibrium condition @xmath123 this condition can be used for simplifying the system of equations which determines the thermonuclear reaction flows .
however , an approximation of this kind is no longer justified , when we consider time scales shorter than the half - lives of the @xmath4-unstable nuclei or temperatures greater than @xmath124=@xmath115 where photodisintegrations become so important that a @xmath2@xmath3-equilibrium is attained rather than a steady flow ( see figure [ fig netzw ] ) .
then , proton - induced reactions can be associated with large quasi - decay constants and , in comparison to @xmath53-decays , small half - lives .
thus , all the abundances of isotonic nuclei which are connected by proton capture reactions can be considered in a quasistatic equilibrium on time scales exceeding the half - lives associated with the quasi - decay constants .
such sequences shall be now defined as quasi - nuclei . in the following paragraphs , the abundances of nuclei in proton capture sequences
shall be calculated under the assumption of a quasi - static equilibrium .
consider the @xmath125 of @xmath113 sequences which consists of @xmath126@xmath127@xmath115 proton captures between @xmath126 nuclei ( _ cf .
_ figure [ fig diagr2 ] : here , two proton capture sequences are shown with @xmath126=@xmath128 and @xmath126=@xmath129 ) .
each nucleus ( @xmath82 ) in a sequence ( @xmath70 ) is involved in @xmath130 reactions including proton captures and photodisintegrations .
furthermore , we assume for simplicity that the abundances of helium and hydrogen stay constant over the time scale of interest and that nucleus @xmath131 is not fed by reactions from another chain @xmath132 . using the equilibrium condition ( [ eq einlpa1c ] )
, the temporal change of the abundance of a nucleus @xmath131 can be written as @xmath133 the terms summed over @xmath71 are leaks out of chain ( @xmath70 ) via nucleus @xmath131 the extreme cases of flux equilibria , like steady flows on the one hand @xmath134 and @xmath2@xmath3-equilibria on the other hand @xmath135 can be discussed on the basis of this equation . here , the general case shall be discussed . for a chain of @xmath126 nuclei ,
@xmath126@xmath127@xmath115 such equations permit to relate the abundance @xmath136 to @xmath122 starting with a `` known '' @xmath137 , the abundance of the last nucleus in a sequence .
thus we obtain for each sequence ( @xmath70 ) an @xmath138@xmath127@xmath139-dimensional system of linear equations which allows us to express each abundance @xmath122 in terms of @xmath137 : @xmath140 the @xmath141 depend only on the constant or slowly varying quasi - decay constants @xmath142 .
the abundance of a sequence can be treated as the abundance of a quasi - nucleus by @xmath143:=@xmath144 .
when @xmath143 is already determined at a point in time @xmath145=@xmath146 , each abundance @xmath122 can be written as @xmath147 with @xmath148=@xmath149 . hence ,
when the abundances of the quasi - nuclei @xmath143 are given at a time @xmath145 , the abundances @xmath150 in each sequence can be determined _
simultaneously_. however , we also have to take care of incoming fluxes into a sequence @xmath151 not jet considered in equation ( [ eq einlpa1 ] ) . as a consequence , the coefficients @xmath141 are no longer constant because they become explicitly abundance - dependent for the following reason : in contrast to the _ outgoing _ fluxes , the _ incoming _ fluxes into a sequence @xmath70 are _ not _ proportional to the abundances @xmath122 of the nuclei in this sequence .
therefore , abundances can no longer be determined simultaneously .
incoming fluxes can nevertheless be considered as long as the abundances do not change too fast during a time step .
we can add to equation ( [ eq einlpa1 ] ) all incoming fluxes from nuclei ( @xmath72 ) of different chains ( @xmath113 ) proportional to their abundances @xmath152@xmath89j ( when no time @xmath145 is indicated , the quantities are given at @xmath145=@xmath146 ) , @xmath153 & & + \sum_{l , m \atop m\neq j } \lambda^{x}_{l , m } y_{l , m}(t_{s-1})=0\ ; .
\label{eq einlpa1d}\end{aligned}\ ] ] this can be solved for @xmath136 , leading to @xmath154 & & -\underbrace{\frac{1}{\lambda^{p,\gamma}_{i-1,j}}\sum_{l , m \atop m\neq j } \lambda^{x}_{l , m } y_{l , m}(t_{s-1 } ) } _ { w_{i-1,j}(\lambda_{l , m}^{x},y_{l , m}(t_{s-1}))}\ , , \end{aligned}\ ] ] or more generally @xmath155 when a slowly varying function @xmath156 is given so that we can approximate @xmath137 by @xmath157 we can extend @xmath158 in ( [ eq einlpa3 ] ) as follows @xmath159 and finally obtain @xmath160 @xmath161 with @xmath162=@xmath163 . setting @xmath164=@xmath165=@xmath166=@xmath165 as initial values , the existence of the functions @xmath162 is guaranteed .
so far we have made two approximations .
( a ) : including reactions within a chain and leaks out of a chain described by equation ( [ eq einlpa1 ] ) . in that case
all abundances could be described simultaneously .
( b ) : when also considering incoming fluxes [ see equation ( [ eq einlpa1d ] ) ] , we become dependent on abundances in other chains . approximating their abundances at @xmath145=@xmath146 by abundances at @xmath167=@xmath168 permits to stay within the same approximation scheme .
but we have to be aware that this is only valid for slowly varying abundances or equivalently small time steps .
we obtained this approximation under the assumption that the abundances of hydrogen and helium stay constant .
@xmath169 and @xmath170 are slowly varying functions in realistic environments .
we can adjust for the change of @xmath169 and @xmath170 and obtain their new values applying relation ( [ eq einlpa2 ] ) @xmath171 as a consequence , the quasi - decay constants must be recalculated after each time step , making use of the updated proton and alpha abundances considering them constant for one time step .
thus it is possible to express simultaneously every abundance @xmath122 in a sequence with the sequence abundance @xmath143 . with an approximation scheme of this kind
, we reduced our full network described by @xmath172 equations to a system of @xmath113 equations . in the following we want to check the accuracy of such an approximation
the calculations were done with a reaction network which consists mainly of proton capture sequences , each starting out from a @xmath173=@xmath174 nucleus .
the number of nuclei in a sequence is determined by the condition that the quasi decay constant @xmath175 of a nucleus @xmath176 into nucleus ( @xmath82+@xmath115 ) is larger than a certain @xmath177 .
when this condition is fulfilled , nucleus ( @xmath82+@xmath115 ) is considered to be in the sequence . for the following calculations @xmath177
was set to @xmath178 s@xmath179 because we wanted to reproduce the energy generation of explosive hydrogen burning on time scales @xmath180=@xmath181 s. as a consequence , a sequence contains not more than 5 nuclei for temperatures @xmath124@xmath116@xmath182 and densities @xmath16@xmath116@xmath17 g @xmath19 .
in addition to the proton capture sequences , @xmath173=@xmath183 nuclei had to be considered in the network .
they are the most neutron - rich nuclei of the hot cno - type cycles , where after a proton capture , the alpha - threshold in the compound nucleus is lower than the proton - threshold , and @xmath31-reactions can occur which compete with the still possible @xmath2-transitions ( see figure [ fig diagr2 ] ) . since the @xmath31- and @xmath2-reactions occur on the same time scale , these @xmath173=@xmath183 nuclei can not affiliated to a proton capture chain but have to be considered separately . between @xmath184c and @xmath185mn , the network can be composed of proton capture reaction sequences and @xmath173=@xmath183 nuclei with one exception . the transition occurs already at compound nucleus @xmath43ne in the ne isotopic chain .
a @xmath31-reaction is possible on the @xmath173=@xmath174 nucleus @xmath49f and a proton capture reaction sequence starts out from @xmath43ne with isospin @xmath173=@xmath186 . the low alpha - threshold of @xmath43ne , which caused an exception of the rule and prevented an ofne - cycle , permits on the other hand the @xmath26o(@xmath187-reaction and triggers the break - out from the cno - cycles .
the network described so far is shown in the figure [ fig diagr1 ] and [ fig diagr2 ] . at temperatures exceeding @xmath124=@xmath188 and @xmath16 @xmath189 @xmath17 g @xmath19 ,
the network must be completed with the @xmath83-reaction .
furthermore , it is very likely that fluxes up to @xmath190kr may influence the energy generation . in the region between @xmath191mn and @xmath190kr ,
the @xmath4-reactions of the for the rp - process relevant nuclei are , with the exception of @xmath192ni , @xmath193ge and @xmath190kr , considerably smaller than the typical time scales of novae and x - ray bursts .
consequently , only the above mentioned nuclei were considered explicitly in the approximation scheme .
the approximated abundances of @xmath194ni and these three waiting point nuclei are compared with full network calculation results in figure [ fig abpa1 ] and [ fig abpa2 ] .
the nuclei used in the approximation scheme are listed in table 1 according to their affiliation to a certain proton capture reaction chain .
nuclei which had to be considered separately ( p , @xmath195he , @xmath173=@xmath183 nuclei , @xmath192ni , @xmath193ge and @xmath190kr ) are given an own sequence number .
the waiting point abundances obtained with this approximation are compared with full network calculation results in figure [ fig abpa1 ] and [ fig abpa2 ] . approximating the network according to section [ approximation ]
, we calculated the energy generation rate @xmath196 and the integrated energy generation @xmath197 for nova - like conditions ( @xmath124=0.2 , @xmath16=@xmath198 g @xmath19 ) and x - ray burst peak conditions ( @xmath124=1.5 , @xmath16=@xmath17 g @xmath19 ) and compared the results to full size network calculations which take into account each nucleus separately . in each case we assumed a solar abundance distribution for the accreted matter .
the results for small temperatures are shown in figure [ fig novae1][fig novae3 ] and those for x - ray burst conditions in figure [ fig xrayb1][fig xrayb3 ] . considering the time scales important for novae ( 1001000 s ) and x - ray bursts ( 110 s ) , we see that the approximation scheme devised in section 3 agrees always to a high degree for the astrophysical applications considered here .
the overall results show only deviations smaller than 15 per cent or even less ( down to 5 per cent ) at a gain in computational speed by a factor of 15 .
several calculations at intermediate temperatures and densities , i. e. @xmath199@xmath200@xmath16@xmath200@xmath17 , @xmath201@xmath200@xmath124@xmath200@xmath182 confirmed the above mentioned efficiency and accuracy of the approximation scheme .
such an approximation would not be very useful if it was only valid for constant temperatures and densities . in order to check its accuracy for varying temperatures and densities we assumed
after peak conditions were obtained in the ignition an exponentially decreasing temperature and an adiabatically expanding accretion layer with a radiation dominated equation of state ( @xmath202 ) : @xmath203 using typical peak temperatures @xmath204 , densities @xmath205 and time scales @xmath180 for x - ray bursts and novae , we again calculated the energy generation ( figure [ fig novaevar1 ] and [ fig xraybvar1 ] ) .
the results show that the approximation is valid for changing temperatures and densities as long as they are in the range of @xmath199@xmath200@xmath16@xmath200@xmath17 , @xmath206@xmath200@xmath124@xmath200@xmath182 .
no significant change in computational speed was observed .
the differences at time scales smaller than @xmath145=@xmath207 s are due to the very fast reaction flows within the proton capture reaction chains : they occur in full scale network calculations until equilibrium is reached according to equation ( [ eq einlpa1c ] ) .
table 1 : + proton capture sequences@xmath208 [ cols= " < , > " , ]
the main motivation for the present investigation was to find a fast and efficient approximation scheme which permits to predict the energy generation in explosive hydrogen burning for a large range of conditions , i.e. temperatures @xmath209@xmath200@xmath124@xmath200@xmath210 and densities @xmath211 g @xmath19@xmath200@xmath16@xmath212 g @xmath19 ) respectively , for applications in hydro calculations which can not afford full nuclear network .
this goal has been achieved .
the energy generation of full network calculations ( 150 nuclei ) could be reproduced with a high accuracy , resulting in deviations of 5 to a maximum of 15 per cent while gaining a factor of about 15 in computational speed .
the additional advantage of this method is that it also permits to predict isotopic abundances with a similar accuracy and can thus even replace full network calculations for this purpose ( see figure [ fig ar33615 ] ) .
we were also able to determine key nuclear properties , which directly enter the precision of calculations for explosive hydrogen burning , its energy generation and abundance determination .
we have identified the even-@xmath30 t@xmath34=1/2 nuclei like @xmath44 mg , @xmath45si , @xmath46s , @xmath47ar and @xmath48ca , as essential targets for proton captures which are in competition with @xmath53-decays . as their small reaction q - values ( less than 2 mev ) do not permit the application of statistical model cross sections , they have to be determined experimentally .
thus , these results are an ideal guidance for future experimental investigations with radioactive ion beams
. nuclei closer to stability permit the application of statistical model cross sections .
nuclei more proton - rich than @xmath213=-1 are only populated for conditions when a @xmath2@xmath3-equilibrium is established .
thus , only their masses , spins and half - lives enter the calculation .
connecting @xmath5-reactions have high enough q - values , even close to the drip - line , that the level densities are sufficient for statistical model cross sections ( this is true at least up to ca and ti , where alpha - induced reactions can play an important role ) .
therefore , the remaining experimental properties which should be determined and enter the calculations in a crucial way are @xmath22-decay half - lives and nuclear masses . the latter control the abundances in a @xmath2@xmath3-equilibrium . beyond se ,
the details of the proton drip - line are actually not that well known yet , but can influence the endpoint of the rp - process .
first calculations which include 2p - capture reactions ( schatz et al .
1996 ) indicate that it seems possible to produce nuclei with a=90 - 100 in the rp - process under x - ray burst conditions , depending on the choice of mass formulae and half - life predictions ( similar to r - process studies ) . especially a region of strong deformation and small @xmath33-capture q - values around @xmath109=80 seems to be predicted with large variations among different mass models , and the even - even @xmath32=@xmath30 nuclei between a=68 and 100 play a dominant role .
the work of f. r. , ch .
f. , and f .- k .
t was supported by the swiss national science foundation grant 20 - 47252.96 .
t. r. acknowledges support by an apart fellowship from the austrian academy of sciences .
h. s. was supported by the german academic exchange service ( daad ) with a `` doktorandenstipendium aus mitteln des zweiten hochschulsonderprogramms '' .
m. w. was supported by the doe grant de - fg02 - 95-er40934 .
arnould , m. , norgaard , h. , thielemann , f .- k . , & hillebrandt , w. 1980 , apj , 237 , 931 ayasli , a. , & joss , p. c. 1982 , apj , 256 , 637 biehle , g. t. 1991 , apj , 380 , 167 biehle , g. t. 1994 , apj , 420 , 364 cannon , r. c. 1993 , mnras , 263 , 817 cannon , r. c. , eggleton , p. p. , zytkow , a. n. , & podsiadlowski , p. 1992 , apj , 386 , 206 champagne , a. , & wiescher , m. 1992 , arn&ps , 42 , 39 coc , a. , mochkovitch , r. , oberto , y. , thibaud , j .- p . , & vangioni - flam , e. 1995 , a&a , 299 , 479 cowan , j. j. , thielemann , f .- k . , & truran , j. w. 1991 , , 208 , 267 fowler , w. a. , caughlan , g. e. , & zimmermann , b. a. 1967 , , 5 , 525 fryer , c. l. , benz , w. , & herant , m. 1996 , , 460 , 801 grres , j. , & wiescher , m. 1995 , , 52 , 412 grres , j. , wiescher , m. , & thielemann , f .- k .
1995 , , 51 , 392 hanawa , t. , sugimoto , d. , & hashimoto , m. 1983 , publ .
japan , 35 , 491 herndl , h. , grres , j. , wiescher , m. , brown , b. a. , & van wormer , l. 1995 , , 52 , 1078 hoffman , r. , woosley , s. e. , fuller , g. m. , & meyer , b. s. 1996 , , 460 , 478 howard , w. m. , meyer , b. s. , & woosley , s. e. 1991 , apjl , 373 , l5 iliadis , c. , grres , j. , ross , j. g. , schaller , k. w. , wiescher , m. , & et al .
1993 , , 559 , 83 kratz , k .-
l . , bitouzet , j .-
p . , thielemann , f .- k . ,
mller , p. , & pfeiffer , b. 1993 , apj , 403 , 216 lewin , w. h. g. , van paradijs , j. , & taam , r. e. 1993 , space sci .
, 62 , 223 meyer , b. s. 1994 , , 32 , 153 rauscher , t. , thielemann , f .- k . , & kratz , k .-
1996 , mem .
, in press rayet , m. , arnould , m. , hashimoto , m. , prantzos , n. , & nomoto , k. 1995 , a&a , 298 , 517 ross , j. g. , grres , j. , iliadis , c. , vonzonkas , s. , wiescher , m. , & et al .
1995 , , 52 , 1681 schatz , h. 1996 , to be published starrfield , s. , truran , j. w. , politano , m. , sparks , w. m. , nofar , i. , & shaviv , g. 1993 , phys . rep . , 227 , 223 sugimoto , d. , & nomoto , k. 1980 , space sci .
, 25 , 155 taam , r. e. , woosley , s. e. , & lamb , d. q. 1996 , apj , 459 , 271 taam , r. e. , woosley , s. e. , weaver , t. a. , & lamb , d. q. 1993 , apj , 413 , 324 thielemann , f .- k . , kratz , k .-
l . , pfeiffer , b. , rauscher , t. , van wormer , l. , & wiescher , m. 1994 , , 570 , 329c van wormer , l. , grres , j. , iliadis , c. , wiescher , m. , & thielemann , f .- k .
1994 , apj , 432 , 326 wallace , r. a. , & woosley , s. e. 1981 , apjs , 45 , 389 wiescher , m. , grres , j. , thielemann , f .- k . , & ritter , h. 1986 , a&a , 160 , 56 woosley , s. e. , taam , r. e. , & weaver , t. a. 1986 , apj , 301 , 601 | hot ( explosive ) hydrogen burning or the rapid proton capture process ( rp - process ) occurs in a number of astrophysical environments .
novae and x - ray bursts are the most prominent ones , but accretion disks around black holes and other sites are candidates as well .
the expensive and often multidimensional hydro calculations for such events require an accurate prediction of the thermonuclear energy generation , while avoiding full nucleosynthesis network calculations . in the present investigation we present an approximation scheme which leads to an accuracy of more than 15 per cent for the energy generation in hot hydrogen burning from @xmath0@xmath1 k , which covers the whole range of all presently known astrophysical sites .
it is based on the concept of slowly varying hydrogen and helium abundances and assumes a kind of local steady flow by requiring that all reactions entering and leaving a nucleus add up to a zero flux .
this scheme can adapt itself automatically and covers situations at low temperatures , characterized by a steady flow of reactions , as well as high temperature regimes where a @xmath2@xmath3-equilibrium is established , while @xmath4-decays or @xmath5-reactions feed the population of the next isotonic line of nuclei .
in addition to a gain of a factor of 15 in computational speed over a full network calculation , and an energy generation accurate to more than 15 per cent , this scheme also allows to predict correctly individual isotopic abundances .
thus , it delivers all features of a full network at a highly reduced cost and can easily be implemented in hydro calculations .
_ subject headings _ : nuclear reactions , nucleosynthesis , abundances stars : novae x - rays : bursts |
radiochromic films , such as gafchromic ebt and ebt 2 films@xcite , are a popular tool to measure dose depositions from ionizing radiation .
most of the previously published studies@xcite have focused on the gafchromic ebt film , the predecessor of the ebt 2 film . however , the active layer of both types of films is the same , which results in comparable outcomes for the ebt and the ebt 2 film @xcite .
the main advantages of these type of films are the fact that these films have the following properties : [ [ section ] ] a near flat response to different photon energies @xcite [ [ section-1 ] ] close to water equivalence ( z@xmath5=6.98 @xcite ) [ [ section-2 ] ] a thickness less then 0.3mm@xcite [ [ section-3 ] ] a low perturbation of the irradiated media , which is certainly important for the evaluation of recent rotating treatment techniques like vmat@xcite [ [ section-4 ] ] a reasonably large dynamic range ( [ 0,10]gy@xcite , [ 0.1,8]gy@xcite ) [ [ section-5 ] ] a high resolution ( 0.22@xmath6m@xcite ) disadvantages of the films are their non - uniformity and production stability , which has an impact on the choice between transmittance ( @xmath0 ) , and optical density ( @xmath7 ) , as dose dependent quantity ( section [ s_odort ] )
. however , the combination of the afore mentioned properties makes gafchromic ebt 2 film , further referred to as film , usable in almost every application in radiotherapy .
for example , at the university hospital leuven the films are clinically used : for in - vivo measurements of total body and electron skin irradiations ; for volumetric modulated arc therapy quality assurance ; for internal audits of brachytherapy , and stereotactic treatments ; for the introduction of new treatment techniques@xcite , and new clinical trials ( flame trial@xcite ) . additionally , there is a large range of research purposes such as ; out of field dosimetry @xcite , estimation of microscopic dose distributions in nano - particle enhanced radiation therapy@xcite , and the optimization of the dosimetry of fluoroscopic ct protocols .
the wide use of these film requires a transparent and accessible absolute calibration protocol .
the calibration methodology aims to use one or two uncut calibration films to represent a whole batch .
the use of an uncut film avoids puzzling with calibration film fragments , and avoids sharp transmission and diffraction edges in the film scanner transmission system .
calibration films and patient quality assurance ( qa ) films are always scanned successively , but they are not necessarily irradiated on the same day .
an error propagation analysis of the calibration protocol is built using the small scale and the large scale variations in the film response .
the calibration protocol is based on eight static fields irradiated on an uncut 8@xmath810 inch film , see figure [ f_calscheme ] .
table [ t_filmsumm ] lists all the films used in this work .
( 203.2,254 ) ( 0,0)(203.2,254 ) ( 0,127)(1,0)203.2 ( 91.6,0)(20,10)[c]gun(101.6,10)(0,1)117.5 ( 101.6,137)(0,1)107 ( 86.6,244)(30,10)[c]target ( 101.6,234)(10,10)[c]x ( 193.2,127)(10,10)[c]y ( 0,239)(1,0)7.5(7.5,239)(10,10)[c ] ( 96.6,127)(10,10)[c]@xmath9 ( 26.6,126.5)(0,1)1 ( 21.6,117)(10,10)[c]@xmath10 ( 76.6,126.5)(0,1)1 ( 71.6,127)(10,10)[c]@xmath11 ( 126.6,126.5)(0,1)1 ( 121.6,117)(10,10)[c]@xmath12 ( 176.6,126.5)(0,1)1 ( 171.6,127)(10,10)[c]@xmath13 ( 101.1,217)(1,0)1 ( 94.1,212)(10,10)[l]@xmath14 ( 101.1,157)(1,0)1 ( 98.6,152)(10,10)[r]@xmath15 ( 101.1,97)(1,0)1 ( 89.1,92)(10,10)[l]@xmath16 ( 101.1,37)(1,0)1 ( 103.6,32)(10,10)[r]@xmath17 ( 148.32,188.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 1 . * ( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath18 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( 48.32,188.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 2 . *
( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath19 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( -1.68,128.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 3 . *
( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath20 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( 98.32,128.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 4 . * ( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath21 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( 148.32,68.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 5 . * ( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath22 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( 48.32,68.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 6 . * ( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath23 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( -1.68,8.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 7 . * ( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath24 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 ( 98.32,8.72 ) ( 56.56,56.56 ) ( 0,0)(56.56,56.56)[lt ] * 8 . *
( 0,28.28)(1,1)28.28 ( 0,28.28)(1,-1)28.28 ( 56.56,28.28)(-1,1)28.28 ( 56.56,28.28)(-1,-1)28.28 ( 14.14,16.5)(28.28,28.28)[cb ] @xmath25 ( 28.28,26)(0,1)4.56 ( 26,28.28)(1,0)4.56 [ cols="^,^,^,<",options="header " , ] for the data sets pooled from a single film , significant red channel @xmath26-estimations were found for film@xmath27 and film@xmath28 , the films with an arbitrary mix of low and high dose values on the different locations . additionally , the blank segment of film@xmath28 resulted in a significant @xmath29-estimation for the green color channel .
the green @xmath30-estimation was borderline significant ( @xmath31 ) . for the calibration sets pooled from two films ,
the @xmath32-estimations become highly significant for the red color channel ( @xmath33 ) .
additionally , a mix of low and high dose values on different locations results in significant estimation of all four @xmath34-parameters ( film@xmath35+film@xmath36 , @xmath37 ) . for this data set , film@xmath35+film@xmath36
, the blue channel @xmath1 could not be determined , @xmath38 and @xmath39 are not significant , this in contrast to @xmath40 . in figure [ f_cal_curve_compar ] the calibration curves resulting from the data sets pooled from film@xmath41 are compared with the calibration curve of film@xmath35+film@xmath36 .
the results are evaluated with a 0.02gy threshold , 1% of 2gy , a commonly prescribed dose .
large differences ( @xmath420.02gy ) between the curves are noticed .
film@xmath28 has the best performance .
film@xmath35 ( increasing mu , with increasing segment number ) under estimates the dose for values larger than 2gy and over estimates the dose between 0.1 and 1.4gy . film@xmath36 ( opposite mu - order of film film@xmath35 ) has an overall underestimation of the dose ( @xmath430gy ) , the low dose range ( values @xmath440.5gy ) has the worst performance .
film@xmath27 ( arbitrary mu - order ) is comparable with film@xmath35 but over estimations are seen where film@xmath35 has underestimations and visa versa .
film@xmath28 ( = film@xmath36 , but with a blank segment and higher dose for segment 6 ) has the better performance for the whole dose range .
+ film@xmath36 is considered as the gold standard , this curve is subtracted from the curves from film@xmath45 , and film@xmath28 .
the 1% dose error , compared to 2gy a common prescribed dose , is illustrated in red.[f_cal_curve_compar],width=264 ] figures [ f_ulocal ] and [ f_uglobal ] show respectively the local and global uniformity .
the local uniformity is smaller than @xmath46 , @xmath47 , and @xmath48 , for the red , green and blue channel .
the global uniformity ranges from 2@xmath49 to @xmath50 .
the blue channel has worse global uniformity than the red and green channel . )
are shown for the following films : a calibration film ( film@xmath36 ) , a uniformly irradiated film ( film@xmath51 ) , and a blank film ( film@xmath52).[f_ulocal],width=264 ] of a uniformly irradiated film ( film@xmath51 ) and three blank films ( film@xmath53).[f_uglobal],width=264 ] the transmittance covariations with the @xmath54-position ( @xmath55 ) are @xmath56 , @xmath57 , and @xmath58 ( red , green , and blue channel ) for the six segment approach .
compared to the eight segment approach , the six segment approach has a 33% reduction of the red channel @xmath55 .
the green and blue channel covariances of the different approaches are comparable .
@xmath59 was not found to be different for the two approaches .
figure [ f_measure8 ] illustrates the dose values based on the calibration curve of the film@xmath35+film@xmath36-data set ( six segment approach ) . the dose from the film measurements , @xmath62 , and the corresponding ic measurements , @xmath63 , are shown in the left figure . on the right side , the residual errors are displayed .
an over estimation is found in the green channel for the un - irradiated segments . using calibration curve based on @xmath64 ( six segment approach ) .
right , dose differences ( @xmath60 ) of the left figure and dose standard deviation ( @xmath65 ) calculated using equations [ e_prop1 ] till [ e_prop4].[f_measure8],width=264 ] in the @xmath66$]gy dose range , excluding all unirradiated segments , the red channel residual errors deviate [email protected]% ( mean@xmath671sd ) from the expected dose , @xmath63 , with a range of @xmath68 $ ] . in the same dose range the green channel residual errors deviate [email protected]% from the expected dose .
these residual errors have a @xmath69 $ ] range , with three outliers of 7.6 , 9.53 , and 16.6% .
the dosimetric accuracy , @xmath65 , is calculated for @xmath70 cm . for the red channel
the calculated dosimetric accuracy is a good estimate for the upper limit of the residual errors . for the green channel
the calculated @xmath65 is an under estimation of this upper limit .
figure [ f_stability ] illustrates an instability for the gafchromic ebt 2-films used in our department .
the @xmath71-values are largely dependent on the production ( variation between different batches ) .
the instability is present in all three color channels but it has different effects , it stresses the necessity to calibrate each batch .
an advantage of our calibration approach is the fact that the estimated calibration parameters indicate the zero dose transmittance ( @xmath1 ) , and therefore , inform the users about the changes of the film .
for the same films , global non - uniformities were noticed . to work with a classic optical density protocol
, a location dependent @xmath72-estimation should be made for each bach ( f.e .
garca - garduo@xcite et al .
proposes an average scan of five randomly selected blank films as @xmath72-estimation for gafchromic ebt films ) . with a throughput of more than one box of films each month
our protocol needs to be as transparent as possible .
therefore , we do not prefer the cumbersome @xmath72-estimation , and @xmath0 is used instead of od .
a stack of films without interleaving tissues allows the films to slide over each other , and static electric charges are created on the film .
the resulting electrostatic forces stick the films to each other . this way the films are only exposed to the environment at the film edges .
the interleaving tissue , avoids the direct contact of the films and serves as a guide for the humidity ( david lewis , isp@xcite , private communication ) .
this is shown in figure [ f_vactest ] , where the films separated by interleaving tissues have similar @xmath1-values , and the films without interleaving tissue have increasing @xmath1-values in function of the location in the stack . when exposed to our local environmental conditions
the films tend to reach a stable state , relative to each other , after 1 day .
meanwhile , the manufacturer reintroduced the interleaving tissue for the transport of the films
. only a correlation between @xmath60 and the @xmath54-location was noticed ( figure [ f_allfit ] ) .
the absence of a correlation between @xmath60 and @xmath73 validates the assumption of using only eight ic measurements to describe the low dose contributions .
the low local uniformity values ( figure [ f_ulocal ] ) support the choice of the ( small ) @xmath74 calibration fields . on the other hand , because of the absence of a @xmath59-difference between the six and the eight segment approach , @xmath59 is not affected by the global non - uniformities .
therefore , the @xmath75 cm segment are small enough to evaluate local uniformity . to cope with global non uniformities a location dependent extension is added to the rational function , additionally the six segment approach was introduced .
this approach decreases the degrees of freedom ( six vs eight data points per film for four @xmath76-parameters ) .
nevertheless , this approach is an improvement for the protocol , indicated by the reduction of @xmath77 , figure [ f_paramsign ] .
figure [ f_cal_curve_compar ] illustrates the better performance of the calibration curve from film@xmath28 .
the improved significance of @xmath34-parameters of film@xmath28 compared to film@xmath27 indicate an improvement of the protocol due to the introduction of a no dose segment .
the necessity of low dose values is further supported by the film@xmath36-results , where the absence of low dose data points in the six segment approach enlarges the residual error , @xmath77 .
this puts the dose levels used in the protocol in question .
future improvements of our protocol are the introduction of 0 mu segment , and the other mu values will be selected to have equal intervals between the @xmath78 data points .
remarkable are the significance levels for the @xmath79-parameters . for the single film calibrations ( film@xmath80 ) ,
only the films with an arbitrary spread of the mus on different locations generate significant results for the red channel . for the two film methodology , the film@xmath35 + film@xmath36-data set , results in significant @xmath81-estimations for _ both _ the red and the green color channel .
this in contrast to the film@xmath82 + film@xmath35-data set , where only the red @xmath32-estimations are significant .
the difference between these two data sets is the mu - order used to irradiate the films .
film@xmath35 + film@xmath36 has an opposite mu - order on the two films while film@xmath82 + film@xmath35 has the same mu - order on the two films .
therefore , the spread of high and low dose segments on different positions is found to be relevant for the calibration of the gafchromic ebt 2-films .
in other words , the calibration curve is location dependent ( @xmath83 ) , which is also mentioned by @xcite@xcite .
the study of such an effect , combining the dose and the x - location ( e.g. a @xmath84-term ) , was outside the scope of is this work .
however , using two films with a spread of different dose levels on different locations , makes our calibration curve an acceptable surrogate for such a location dependent calibration curve .
although , @xmath0 is used instead of @xmath7 , it is interesting to compare the uniformity results with literature values .
after all , both @xmath0 and @xmath7 form the input of a calibration curve , and the error on these input will determine the error of the resulting dose values .
excellent local uniformity values were found ( @xmath85 ) which is lower than @xmath86 and @xmath87 relative standard deviations for the md-55 - 2-film listed in the tg 55 report @xcite .
zhu et al .
@xcite reports @xmath50-@xmath88 global od - variations for the md-55 film , which is comparable with our transmittance results ( @xmath89 ) .
nevertheless , adaptations were necessary to cope with the global non uniformity ( six segment approach , and @xmath54-dependent parameter in the calibration curve ) . in a valid calibration protocol @xmath60
should have a normal distribution with a zero mean and a standard deviation from the previous section ( @xmath90 ) .
this relation is not evaluated because of the small amount of data , @xmath91 per dose level .
therefore , we have chosen an over estimation of @xmath65 , introduced by using the _ maximum _ of the local and global uniformity variations in equations [ e_sigmat ] and [ e_covtx ] .
the red color channel @xmath65-prediction was able to estimate the maximal dose errors , figure [ f_measure8 ] .
for the green channel the calculated @xmath65-values are not sufficient to estimate the maximal dose errors in the protocol .
neither was the protocol capable to convert @xmath92 to dose , see figure [ f_measure8 ] .
to cope with this problem , blank segments could be introduced in the calibration protocol .
the significance of the @xmath93 and @xmath94-estimations for the film with a blank segment ( film@xmath28 ) supports this assumption .
the proposed calibration protocol for absolute gafchromic film dosimetry has a straightforward association between transmittance and dose , equation [ e_itodose2 ] .
the approachable protocol required only eight preparation measurements , and a single reference measurements on the day of calibration .
the use of simple static 4@xmath84 fields is suited for this purpose even as the use of ic - measurements to map the low dose contributions .
film changes are reported both , due to the production process as due to the storage environment .
the strength of our protocol is that each calibration characterizes the physical parameters of the films , @xmath1 , @xmath2 , and a factor scaling the impact of the dose ( @xmath95 ) .
however , because of non - uniformities the original intended association between @xmath0 and @xmath96 required an adaptation .
an additional term ( @xmath97 ) and the avoidance of certain segments , are positively evaluated as improvements of the protocol .
the protocol requires a spread of low and high dose segments on two calibration films and the segments with a higher spread of the residual dose errors should be excluded .
additionally the use of a blank segments is advisable .
all color channels are dose dependent , but a trustworthy calibration protocol with error prediction was only found for the red channel in the @xmath980,2.5]$]gy dose range . in this dose range
the red channel dose error range equals @xmath68 $ ] . for the red color channel an upper limit for the dose error
could be predicted using an error propagation analysis .
the green color channel has the higher residual errors , and requires more attention to the estimation of @xmath1 .
the blue color channel is dose dependent but could not be calibrated with the proposed protocol
. + [ [ preparation - measurements ] ] preparation measurements + + + + + + + + + + + + + + + + + + + + + + + + eight preparation measurements are performed once to estimate the low dose contributions .
[ [ storage-1 ] ] storage + + + + + + + the films are stored in separate folders of a file cabinet to ensure the same environmental exposure off all films .
[ [ blank - check ] ] blank check + + + + + + + + + + + after a few days of storage , for stabilization purposes , all films are scanned unirradiated .
thereafter , an automated script calculates @xmath99 , as well as the local uniformity off all the films .
[ [ dose - levels ] ] dose levels + + + + + + + + + + + a blank segment is introduced for a better estimation of @xmath100 . in our current protocol the dose range is extended from 0 to @xmath101 @xmath102gy , using mu ranging from @xmath103 to @xmath104 . where the number of mus are chosen to be equally distributed in the red transmittance domain . [ [ films ] ] 2 films + + + + + + + two calibration films
are irradiated using opposite mu - orders .
[ [ calibration - check ] ] calibration check + + + + + + + + + + + + + + + + + before using clinical films , the calibration films are scanned and evaluated .
[ [ segment - selection ] ] segment selection + + + + + + + + + + + + + + + + + based on the blank check and the calibration check a decision is made wether certain segments should be avoided or not ( e.g. segments with @xmath105 cm ) . in our latest clinical evaluations such a segment avoidance was not necessary ( batch nr .
a09031001b , a11011001 , a12171002b ) .
[ [ h - post - irradiation ] ] 48h post irradiation + + + + + + + + + + + + + + + + + + + + we wait 48 hours before scanning the films . [ [ scan - all - films - in - one - session ] ] scan all films in one session + + + + + + + + + + + + + + + + + + + + + + + + + + + + + all the films that need to be evaluated are scanned successively , including the calibration films .
[ [ convert - to - dose ] ] convert to dose + + + + + + + + + + + + + + + a calibration curve is created for the scan session and all the films are converted to dose .
[ [ calibrate - each - box - of - films ] ] calibrate each box of films + + + + + + + + + + + + + + + + + + + + + + + + + + + [ [ acknowledgements ] ] * acknowledgements * + + + + + + + + + + + + + + + + + + mr .
k. poels is acknowledged for extensive discussions .
the authors like to thank isp and peo for their fruitful discussion and their support in this work .
equation [ e_itodose ] ( [ e_a1 ] ) can be transformed to [ e_dosetoi ] ( [ e_a2 ] ) using the following routine juggling .
these equations can be reduced further using three physical parameters .
the first two parameters characterizes the film using the zero and infinite dose transmissions ( @xmath1 and @xmath107 ) .
the third parameter , @xmath3 scales the impact of the dose .
24ifxundefined [ 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 _ _ , edited by and ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | * purpose : * to relate the physical transmittance parameters of the water equivalent gafchromic ebt 2 film with the delivered dose in a transparent absolute calibration protocol . the protocol should be easy to understand , easy to perform , and should be able to predict the residual dose error .
* methods : * the protocol uses at most two uncut calibration films for a whole batch .
the calibration films are irradiated on different positions with static fields with varying dose levels .
the transmittance trough the film , @xmath0 , is preferred over the optical density .
an analytical function is used to associate @xmath0 with the delivered dose based on physical characteristics of the film : the minimal and maximal @xmath0-values , @xmath1 and @xmath2 , and a factor scaling the dose , @xmath3 .
the dose uncertainty of the protocol is calculated using an error propagation analysis of the calibration curve .
this analysis makes use of the small scale and the large scale variations in the film response ( local and global uniformity ) . before calibrating the film ,
the stability of unirradiated films is evaluated .
the relevance of the positions of the different dose levels is studied .
* results : * both a production and a spatial dependency of the unirradiated films was noticed .
this spatial dependency required an avoidance of a quarter of the film .
the transmittance is also affected by the storage and the transport conditions .
the lowest residual errors and the highest significance for @xmath1 , @xmath2 , and @xmath3 was achieved with arbitrary dose levels for different positions on the film .
two calibration films give the better results .
significant calibration curves were found for both the red and the green color channel .
large differences ( @xmath4gy ) between calibration curves , with different positions for the dose levels are seen . in the [ 0.04,2.5]gy dose range the red channel calibration curve has dose errors ranging from -2.3% to 4.9% . for the green channel
the error range is [ -3.6%,6.9% ] .
the calculated error propagation was able to predict an upper limit for the red channel dose errors .
for the green channel this prediction was not achieved .
* conclusions : * the gafchromic ebt2 films are properly calibrated with an accessible robust calibration protocol .
the protocol largely deals with the uniformity problems of the film .
the proposed method allowed to relate the dose with the red channel transmittance using only @xmath1 , @xmath2 , and a dose scaling factor . based on the local and global uniformity the red channel dose errors could be predicted to be smaller than 5% . |
let @xmath8 be a simple graph of order @xmath9 .
we use the the following notations : the set of vertices adjacent in @xmath1 to a vertex of a vertex subset @xmath10 is the open neighborhood @xmath11 of @xmath12 .
the closed neighborhood @xmath13 $ ] also includes all vertices of @xmath12 itself . in case of a singleton set @xmath14
we write @xmath15 and @xmath16 $ ] instead of @xmath17 and @xmath18 $ ] , respectively .
@xmath19 denotes the automorphism group of @xmath1 .
a labeling of @xmath1 , @xmath20 , is said to be @xmath21-distinguishing , if no non - trivial automorphism of @xmath1 preserves all of the vertex labels .
the point of the labels on the vertices is to destroy the symmetries of the graph , that is , to make the automorphism group of the labeled graph trivial .
formally , @xmath22 is @xmath21-distinguishing if for every non - trivial @xmath23 , there exists @xmath24 in @xmath25 such that @xmath26 .
the distinguishing number of a graph @xmath1 is defined by @xmath27 this number has defined by albertson and collins @xcite .
if a graph has no nontrivial automorphisms , its distinguishing number is one . in other words , @xmath28 for the asymmetric graphs .
the other extreme , @xmath29 , occurs if and only if @xmath30 . the distinguishing number of some examples of graphs , @xmath31 for every @xmath32 , and @xmath33 for @xmath34 , @xmath35 for @xmath36 .
also @xmath37 , for @xmath38 , and @xmath39 , for @xmath40 .
authors in @xcite have shown that removing a vertex of @xmath1 can decrease the distinguishing number by at most one but can increase by at most to double of distinguishing number of @xmath1 . also for each connected graph @xmath1 and @xmath41 , @xmath42 .
a domination - critical ( domination - super critical , respectively ) vertex in a graph @xmath1 is a vertex whose removal decreases ( increases , respectively ) the domination number .
bauer et al .
@xcite introduced the concept of domination stability in graphs .
the domination stability , or just @xmath43-stability , of a graph @xmath1 is the minimum number of vertices whose removal changes the domination number .
motivated by domination stability , we introduce the distinguishing stability of a graph .
let @xmath1 be a graph of order @xmath9 .
the stabilizing on the distinguishing number , or just distinguishing stability , @xmath3 of graph @xmath1 is the minimum number of vertices whose removal changes the distinguishing number . also we introduce and study the edge distinguishing stability number ( distinguishing bondage number ) of @xmath1 and compute edge distinguishing stability of some specific graphs . in the next section
we compute the distinguishing stability of some specific graphs .
we obtain general bounds , and a relationships between the distinguishing stabilities of @xmath1 and @xmath5 , where @xmath5 denotes the graph obtained from @xmath1 by removal of a vertex @xmath44 and all edges incident to @xmath44 , in section 3 . finally we consider and study the edge distinguishing stability number of graphs in section 4 .
in this section , first we compute the distinguishing stability of some specific graphs .
we start with paths and cycles .
a path is a connected graph in which two vertices have degree one and the remaining vertices have degree two .
let @xmath45 be the path with @xmath46 vertices as shown in figure [ fpath ] .
a path @xmath45 with vertex set @xmath47.,scaledwidth=40.0% ] [ path ] for any @xmath48 , the distinguishing stability of @xmath45 , is @xmath49 , while @xmath50 , @xmath51 , @xmath52 , and @xmath53 .
* proof .
* it is clear that @xmath54 for @xmath48 . on the other hand by removing the third and sixth vertex of @xmath45 , i.e. , vertices labeled by numbers @xmath55 and @xmath56 , the graph @xmath45 convert to disjoint union of the two paths @xmath57 and one path @xmath58 , and so @xmath59 . since @xmath31
, so @xmath49 for @xmath48.@xmath60 the following proposition obtain immediately from proposition [ path ] . for any @xmath61 ,
the distinguishing stability of @xmath62 is , @xmath63 , while @xmath64 , and @xmath65 . with respect to the value of the distinguishing number of the complete graphs and complete bipartite graphs we can prove the following proposition : 1 . for any @xmath66 , @xmath67 .
2 . for any @xmath68 , @xmath69 , except , @xmath70 .
the @xmath46-book graph @xmath71 ( figure [ book ] ) is defined as the cartesian product @xmath72 .
we call every @xmath73 in the book graph @xmath74 , a page of @xmath74 .
all pages in @xmath74 have a common side @xmath75 .
we shall compute the distinguishing stability number of @xmath74 .
the following theorem gives the distinguishing number of the book graph .
book graph @xmath74.,scaledwidth=50.0% ] @xcite the distinguishing number of @xmath74 @xmath76 is @xmath77 .
the distinguishing stability of the book graph is @xmath78 * proof .
* by removing the two central vertices of @xmath74 , the book graph @xmath74 convert to disjoint union of @xmath46 paths @xmath57 , denoted by @xmath79 .
it can be computed that the distinguishing number of @xmath79 is @xmath80 which is different from the distinguishing number of @xmath74 , @xmath77 .
so @xmath81 . if the value of @xmath82 is square , then @xmath83 , and so by removing a noncentral vertex of @xmath74 , say @xmath44 , we have a book graph @xmath84 and the path @xmath85 , that start point of @xmath85 is identified with one of the central point of @xmath84 , and so @xmath85 is fixed by any automorphism of @xmath86 .
so @xmath87 , and hence @xmath88.@xmath60 the friendship graph @xmath89 @xmath71 can be constructed by joining @xmath46 copies of the cycle graph @xmath90 with a common vertex . to compute the distinguishing stability of @xmath89 , we need the following theorem .
@xcite[distfan ] the distinguishing number of the friendship graph @xmath89 @xmath76 is @xmath91 [ pro fan ] the distinguishing stability of the friendship graph is @xmath92 * proof .
* set @xmath93 and consider the graph @xmath89 as shown in figure [ friendsep ] .
the graph @xmath94 is two graphs @xmath95 and @xmath96 such that their central vertices are identified . since @xmath97 is the minimum number which @xmath98 is square , so @xmath99 . on the other hand , since @xmath96 and @xmath95 are two nonisomorphic graphs , so @xmath100 , hence @xmath101 .
thus @xmath102 .
now we show that @xmath103 . first note that if we remove the central vertex of @xmath89 , say @xmath104 , then the value of the distinguishing number of @xmath105 is equal @xmath106 .
in general , if @xmath107 are @xmath108 noncentral vertices of @xmath89 , then @xmath109 , because the central vertex @xmath104 is fixed under each automorphism and @xmath110 . by using above point and regarding to the value of @xmath97 ,
if we remove less than @xmath97 vertices of @xmath89 , say @xmath111 , then @xmath112 , and so @xmath113 . therefore @xmath114 .
@xmath60 friendship graph @xmath89.,scaledwidth=50.0% ] with respect to the value of @xmath115 for different values of @xmath46 we can prove the following result : for every natural number @xmath116 , there exists a graph @xmath1 such that @xmath117 .
* proof . * if @xmath118 , then it is suffice to consider the complete graphs . for @xmath119
the result follows immediately from proposition [ pro fan]@xmath60
in this section , we study the relationship between the distinguishing stabilities of graphs @xmath1 and @xmath5 , where @xmath7 .
also we obtain upper bounds for @xmath3 .
[ minus ] let @xmath1 be a graph and @xmath44 be a vertex of @xmath1 , then @xmath120 * proof . * if @xmath121 , then @xmath6 . if @xmath122 , then @xmath123 , and so we have the result.@xmath60 using proposition [ minus ] and mathematical induction , we have @xmath124 where @xmath125 and @xmath126 . only using this formula we can get different upper bounds for @xmath3 of graph @xmath1 .
we state some these upper bounds in the following theorem which for the prove of each case , it is sufficient to remove the vertices of @xmath1 until the induced subgraph which stated in the hypothesis , appears . next using @xmath124 and the value of @xmath127 , we can have the result .
let @xmath1 be a simple graph of order @xmath9 .
* @xmath128 . *
if @xmath129 is maximum degree of graph @xmath1 , and @xmath1 has the star graph @xmath130 as the induced subgraph with @xmath131 , then @xmath132 . *
if @xmath2 is the diameter of graph @xmath1 , and @xmath1 has the path @xmath133 as the induced subgraph , then @xmath134 * @xmath135 , where @xmath136 is the clique number of @xmath1 .
let @xmath1 be a graph of order @xmath46 , then @xmath137 .
* proof .
* let @xmath117 .
so for every @xmath138 ( @xmath139 ) vertices of @xmath1 , say @xmath140 , we have @xmath141 . since the distinguishing number of a graph is at most equal to its order
, so @xmath142 .
@xmath60 a graph and its complement , always have the same automorphism group while their graph structure usually differs .
hence @xmath143 for every simple graph @xmath1 . in the following theorem
we use this equality to show that the distinguishing stability of a graph and its complement are the same .
[ complement ] let @xmath1 be a simple graph , then @xmath144 . * proof .
* since @xmath143 and @xmath145 , we have the result .
@xmath60 [ dbound ] if there exists a vertex @xmath44 of @xmath1 such that @xmath146 , then @xmath147 .
* proof . * if @xmath122 , then @xmath148 , and so @xmath147 . if @xmath149 , then we use mathematical induction on the order of @xmath1 .
it can be seen that the result is true for small value of @xmath46 .
let @xmath147 for all graphs of order @xmath150 .
suppose that @xmath151 , in this case by induction hypothesis we have @xmath152 .
thus we can write @xmath153 therefore the result follows.@xmath60 we think that for any simple graph @xmath1 the inequality @xmath154 is true . however , until now all attempts to find a proof failed .
so we propose the following conjecture here .
let @xmath1 be a simple connected graph , then @xmath154 .
let @xmath44 be a vertex in @xmath1 .
the contraction of @xmath44 in @xmath1 denoted by @xmath155 is the graph obtained by deleting @xmath44 and putting a clique on the ( open ) neighbourhood of @xmath44 .
note that this operation does not create parallel edges ; if two neighbours of @xmath44 are already adjacent , then they remain simply adjacent ( see @xcite ) . in the end of this section
, we study the distinguishing stabilities of graph @xmath156 .
let @xmath1 be a graph and @xmath44 be a vertex of it , and let @xmath157 be the added edges to the neighbours of @xmath44 in the construction of @xmath155 .
suppose that @xmath158 are all different neighbours of @xmath44 which are incident to at least one of the edges @xmath157 .
we have @xmath159 * proof . * first of all note that if @xmath160 is an edge incident to the vertex @xmath104 of @xmath1 , then by proposition [ minus ] , @xmath161 .
using theorem [ complement ] we can write @xmath162 by the first sentence of proof , it can be concluded that @xmath163 now using theorem [ complement ] , we obtain the result.@xmath60
in this section we study the edge stability ( bondage ) number on the distinguishing number of a graph . the bondage number on the distinguishing number @xmath164 , ( or the edge distinguishing stability number ) is the minimum number of edges whose removal changes the distinguishing number .
first we determine @xmath164 for several families of graphs including complete ( bipartite ) graphs , cycles , and paths .
* proof . * with respect to theorem [ distfan ] , we know that for every @xmath169 there exists @xmath176 such that @xmath177 .
also it can be concluded that there exist @xmath178 consecutive friendship graphs with the distinguishing number @xmath179 . without loss of generality ,
we assume that @xmath180 .
we claim that \(i ) if we remove the edges @xmath189 from @xmath190 , then @xmath190 converted to the friendship graph @xmath191 such that the end vertex of @xmath192 paths of order three identified to the central vertex of @xmath191 .
it can be computed that @xmath193 , and hence @xmath194 . if we remove less than @xmath192 edges from @xmath190 , say @xmath195 , @xmath196 , then @xmath197 has @xmath198 where @xmath199 as its fixed induced subgraph , and hence @xmath200 , so @xmath201 .
\(ii ) if we remove the edges @xmath202 from @xmath203 , then we obtain a graph such that it has made by identifying the central vertices of the star graph @xmath204 and @xmath205 , and so @xmath206 . hence @xmath207 .
if we remove less than @xmath208 edges from @xmath203 , say @xmath195 , @xmath209 , then @xmath210 has @xmath211 where @xmath212 as its fixed induced subgraph , and hence @xmath200
. therefore @xmath213 . to prove the second part of theorem , it is sufficient to note that there exists @xmath214 such that @xmath215 . by a similar argument we can show that @xmath216 , and @xmath217 for @xmath182 .
so we have the result .
@xmath60 * proof . * using @xmath218 and mathematical induction , we have @xmath227 where @xmath228 and @xmath229 . to prove each case , it is sufficient to remove the edges of @xmath1 until the connected components of size greater than or equal one has been stated in proof of each parts , appears .
next using @xmath227 and the value of @xmath230 , we can have the result . 1
. the connected component of size greater than or equal one is the path @xmath57 , so @xmath231 . since @xmath232 so @xmath220 .
2 . the connected component of size greater than or equal one is @xmath130 , so @xmath233 . since @xmath234 , the result follows .
the connected component of size greater than or equal one is @xmath133 , so @xmath235 . if @xmath222 , then @xmath236 , and if @xmath224 , then @xmath237 .
hence the result follows .
the connected component of size greater than or equal one is @xmath238 , so @xmath239 . if @xmath240 , then @xmath241 , and if @xmath242 , then @xmath243 .
hence the result follows .
@xmath60 in 1956 , nordhaus and gaddum obtained the lower and upper bounds for the sum of the chromatic numbers of a graph and its complement ( actually , the upper bound was first proved by zykov @xcite in 1949 ) .
since then , nordhaus - gaddum type bounds were obtained for many graph invariants .
an exhaustive survey is given in @xcite .
here , we give nordhaus - gaddum type inequalities for the bondage number on the distinguishing number of a graph .
* proof .
* the proof is by induction on @xmath219 , the number of edges of @xmath1 .
it can easily be seen that it is true for small value of @xmath219 .
so suppose that the result is true for graphs with size less than @xmath219 .
by @xmath218 , we can obtain @xmath247 , and so using @xmath248 and induction hypothesis we get the result.@xmath60 note that the upper bound of theorem [ nordbandag ] is sharp and is achieved for example by the family of cycle graphs @xmath62 with @xmath61 .
in fact , if @xmath250 are consecutive vertices of @xmath62 , then @xmath251 is an asymmetric graph , and so @xmath252 . about
the sharpness of lower bound , it is sufficient to consider the complete graphs .
* proof .
* if there exists an edge @xmath160 such that @xmath255 , then @xmath256 , and so we have the result . if @xmath257 for all edges of @xmath1 , then by induction on the size of @xmath1 , we can write @xmath258 where @xmath160 is incident to one of the vertices of @xmath1 such that their removal change the distinguishing number .
hence it can be concluded that @xmath259 for such edges . @xmath60 | the distinguishing number @xmath0 of a graph @xmath1 is the least integer @xmath2 such that @xmath1 has a vertex labeling with @xmath2 labels that is preserved only by a trivial automorphism .
the distinguishing stability , of a graph @xmath1 is denoted by @xmath3 and is the minimum number of vertices whose removal changes the distinguishing number .
we obtain a general upper bound @xmath4 , and a relationships between the distinguishing stabilities of graphs @xmath1 and @xmath5 , i.e. , @xmath6 , where @xmath7 .
also we study the edge distinguishing stability number ( distinguishing bondage number ) of @xmath1 .
department of mathematics , yazd university , 89195 - 741 , yazd , iran + [email protected] , [email protected] * keywords : * distinguishing number ; stability ; bondage number * ams subj . class . * : 05c25 |
in superconducting rings , the requirement of a single - valued , macroscopic quantum mechanical wave - function , combined with the intimate relation between the quantum mechanical phase and the vector potential , results in flux quantization.@xcite under the appropriate conditions , two of the flux quantized states can become degenerate , characterized by time - reversed , persistent supercurrents circulating in a clockwise and counter - clockwise direction .
these macroscopic circulating currents can be thought of as an analog to an electronic spin . in pioneering work
, davidovic et al.@xcite showed that arrays of superconducting rings can be used as models for spin systems , since they interact anti - ferromagnetically upon cooling .
however , the davidovic arrays never showed long - range anti - ferromagnetic ordering .
they speculated that one reason for this lack of ordering is that , for there to be two degenerate states , such rings must be cooled in a field equivalent to a half - integer multiple of the superconducting flux quantum @xmath5 per ring .
therefore variations in the lithographically patterned areas of these rings result in different fluxes through different rings in the same field , causing disorder .
superconducting @xmath3-rings have an intrinsic phase shift of @xmath3 in the absence of an externally applied field or supercurrent .
@xcite such rings have a doubly degenerate ground state in zero applied flux , should not have this source of disorder , and may therefore be a more ideal model for a spin system
. however , use of @xmath3-rings for a model spin system requires a large number of rings . the first superconducting @xmath3-rings,@xcite which depended on the momentum dependence of the pairing wavefunction in the high - t@xmath6 cuprate perovskite superconductors ,
were made using technologies which would be difficult to extend to many devices .
recently a technology that allows for photolithographic fabrication of @xmath3-shift devices and arrays of great complexity , using yba@xmath0cu@xmath1o@xmath2-au - nb ( ybco - au - nb ) ramp - edge tunneling junctions has been demonstrated.@xcite moreover , it has recently been demonstrated that @xmath3-rings can also be fabricated using josephson junctions with ferromagnetic layers in the tunnel barriers .
@xcite in this paper we report on experiments in which the ybco ramp - edge technology was used to fabricate large arrays of @xmath3-rings . a first report on work with similar arrays appeared in ref . .
the arrays were cooled in various magnetic fields , and the final spin " states of the arrays were determined with scanning squid microscopy .
long range anti - ferromagnetic ( afm ) ordering was observed in the 1d arrays . although stronger anti - ferromagnetic correlations were observed in our 2d @xmath3-ring arrays than were reported previously for the conventional ( @xmath7-ring ) arrays , in neither case did afm ordering extend beyond a few lattice sites .
although using @xmath3-rings eliminated one source of disorder , inhomogeneous flux biasing due to the applied fields required for @xmath7-rings , there are other sources of disorder , including inhomogeneities in the ring critical currents and critical temperatures .
we will discuss the influences of these sources on our results , as well as dynamic effects , using monte - carlo modelling of the cooling process .
various samples consisting of one- and two - dimensional @xmath3-ring arrays have been realized using ramp - type yba@xmath0cu@xmath1o@xmath2 - au - nb josephson contacts .
the fabrication of ramp - type yba@xmath0cu@xmath1o@xmath2 - au - nb josephson has been described previously in detail in ref.s . in short
, the samples were prepared by first epitaxially growing a bilayer of [ 001]-oriented yba@xmath0cu@xmath1o@xmath2 and srtio@xmath1 by pulsed - laser deposition on [ 001]-oriented srtio@xmath1 single crystal substrates . for
the 1d array samples a 150 nm yba@xmath0cu@xmath1o@xmath2 and a 100 nm srtio@xmath1 film were used , while for the 2d array samples a 340 nm yba@xmath0cu@xmath1o@xmath2 and a 67 nm srtio@xmath1 film were used . in these bilayers the basic layout of the structures , which will be described below in more detail ,
is defined by photolithography and ar ion milling .
this process results in interfaces with a slope of @xmath8 with respect to the substrate , which provides access to the @xmath9-planes of the yba@xmath0cu@xmath1o@xmath2 and allows the exploitation of @xmath10-wave phase effects .
special care is taken to align all interfaces accurately along one of the yba@xmath0cu@xmath1o@xmath2 [ 100 ] axes .
after etching the ramp and cleaning of the sample , a 7 nm yba@xmath0cu@xmath1o@xmath2 interlayer is deposited , the function and properties of which are described in ref . , followed by the in - situ pulsed - laser deposition of a au barrier - layer of 12 nm for the 1d array samples and 20 nm for the 2d array samples . a 160 nm nb counter electrode is then formed by dc sputtering and structured by lift - off .
each chip contained several reference - junctions . at t = 4.2 k
, these showed a typical critical current per micrometer junction - width @xmath11 ma/@xmath12 m for the 1d array samples and @xmath13 ma/@xmath12 m for the 2d array samples . from these , a value for the josephson penetration depth @xmath14 m
( t = 4.2 k ) for all samples is deduced , which is the characteristic length scale over which josephson vortices ( fluxons ) extend .
-rings used for the 2d arrays reported in this paper . type ( a ) was used in the square arrays , and type ( b ) was used in the triangular , honeycomb , and kagom arrays.,width=336 ] the sample magnetic fields were imaged with a high resolution scanning squid microscope.@xcite the squid microscope images shown here were made at a temperature of t @xmath15 5k , with the sample cooled and imaged in the same fields .
the size of the pickup loop used will be indicated for each image .
the samples were warmed through the superconducting critical temperature of the nb and cooled at controlled rates either using a non - inductive heater , or by passing warm @xmath16he gas past the sample .
-rings , with 2.7@xmath12 m junctions and 11.5@xmath12 m ring to ring spacings.,width=336 ] a first configuration for which the generation and coupling of half - integer flux quanta was investigated is the zigzag array,@xcite several instances of which are shown schematically in figure [ fig : hilgfacets ] . in these structures ,
the @xmath10-wave order parameter of the yba@xmath0cu@xmath1o@xmath2 induces a difference of @xmath3 in the josephson phase - shift @xmath17 across the yba@xmath0cu@xmath1o@xmath2-au - nb barrier for neighboring facets . for facet lengths
@xmath18 in the wide limit , i.e. , @xmath19 , the lowest - energy ground state of the system is expected to be characterized by a spontaneous generation of a half - integer flux - quantum at each corner .
this half - fluxon provides a further @xmath3-phase change between neighboring facets , either adding or subtracting to the @xmath10-wave induced @xmath3-phase shift , depending on the half - flux quantum polarity . in both cases
this leads to a lowering of the josephson coupling energy across the barrier , as this energy is proportional to ( 1 - cos@xmath17 ) .
we studied three types of facetted junctions .
the first ( fig .
[ fig : hilgfacets]a ) was an isolated , continuous junction with many adjacent facets .
the second ( fig .
[ fig : hilgfacets]b ) was lithographically patterned to electrically isolate each half - fluxon from its neighbor
. the final type ( fig .
[ fig : hilgfacets]c ) had two continuous facetted junctions close together , but electrically isolated from one another , to test for field coupling between 1d arrays of half - fluxons .
height ybco ( nm ) & 340 & 340 & 340 & 340 + height sto ( nm ) & 67 & 67 & 67 & 67 + height nb ( nm ) & 160 & 160 & 160 & 160 + width jj 1 ( @xmath12m)@xmath20 & 2.75 & 2.70 & 2.70 & 2.70 + width jj 2 ( @xmath12m)@xmath20 & 2.75 & 3.75 & 3.75 & 3.75 + hole ( @xmath12 m ) & 2.50 & @xmath213.10 & @xmath213.10 & @xmath213.10 + self - inductance ( ph)@xmath22 & 3.93 & @xmath213.90 & @xmath213.90 & @xmath213.90 + self - inductance ( ph)@xmath23 & 4.57 & - & - & - + nearest neighbor & 11.5 & 11.5 & 11.5 & 11.5 + distance ( @xmath12 m ) & & & & + nearest neighbor & 0.025 & - & - & - + mutual ( ph)@xmath23 & & & & + + + + the 2-dimensional @xmath3-ring arrays were made up of individual rings patterned as indicated schematically in figure [ fig : ec1schem ] .
there were two types of rings , square ( fig .
[ fig : ec1schem]a ) and hexagonal ( fig .
[ fig : ec1schem]b ) .
the rings were patterned into arrays with 4 different geometries , as shown in the scanning electron microscopy images of figure [ fig : ec1sems ] . in all cases
the nearest - neighbor distances in these arrays was 11.5@xmath12 m center to center .
the details of the ring geometries , critical currents , self - inductances , and mutual - inductances are given in table 1 .
junctions , cooled in nominally zero field , and imaged with a 4@xmath12 m diameter pickup loop .
( a ) continuous junction with 40@xmath12 m facet lengths .
( b ) electrically disconnected junction with 40@xmath12 m between facet corners .
( c ) electrically disconnected junction with 20@xmath12 m between facet corners .
two parallel facetted @xmath24 junctions with 40@xmath12 m ( d ) , 20@xmath12 m ( e ) , and 10@xmath12 m ( f ) between facet corners .
the apparent curvature of the junctions in these images is an artifact of the scanning mechanism.,width=336 ] junction with 10 facets each 40@xmath12 m long , cooled in fields of 0nt ( a ) , 32nt ( b ) , 74nt ( c ) , and 110nt ( d ) , and imaged at 4.8k with an 8@xmath12 m square pickup loop.,width=336 ]
figure [ fig : hilgquad ] shows representative scanning squid microscope images of 6 zigzag @xmath24 facet junctions .
all of these images were of samples on the same substrate , and imaged in the same cooldown in nominal zero field ( @xmath25 t ) at t = 4.2k .
in all three types of facetted junctions , half - fluxon josephson vortices were spontaneously generated at the points where the facets met , as expected . when the connected 1d arrays ( fig .
[ fig : hilgfacets]a ) were cooled in zero field through the niobium superconducting transition temperature , the signs of the half - fluxons strongly tended towards perfect anti - ferromagnetic ordering , in which the persistent supercurrents alternated between clock - wise and counter clock - wise flow ( fig .
[ fig : hilgquad]a , d , e , f ) .
there are two possible mechanisms for this ordering . in the first ,
ordering is driven by the minimization of the total junction josephson coupling energy during the cooling process .
we refer to this as phase " coupling , because the josephson currents and energies are determined by the difference in superconducting phase across the junction @xmath17 . in the second , the anti - ferromagnetic
ordering is driven by a minimization of the total magnetic field energy in the junction and its environment .
we refer to this as field " coupling . in order to determine the relative strengths of these mechanisms , facetted junctions
were fabricated with each half - fluxon electrically isolated from its neighbor ( fig .
[ fig : hilgfacets]b ) .
this should eliminate the phase coupling mechanism .
indeed , when this is done the anti - ferromagnetic ordering in these 1d arrays is much weaker .
two examples are shown in fig .
[ fig : hilgquad]b , c . in fig .
[ fig : hilgquad]b the half - fluxons all have the same orientation .
we believe that this is because they align with a small residual field .
[ fig : hilgquad]c shows a more random alignment of the isolated half - fluxons .
-rings , in the geometries illustrated in fig .
[ fig : ec1sems ] , with values for the ring parameters given in table i. these images were taken at 4.2k with a 4@xmath12 m diameter pickup loop after cooling in nominally zero field at 1 - 10mk / sec .
the spin up fractions @xmath26 and full scale variations in the scanned squid sensor flux ( @xmath27 ) were @xmath26=0.56 , @xmath28 for the square lattice ; @xmath26=0.51 , @xmath29 for the honeycomb lattice ; @xmath26=0.50 , @xmath30 for the triangular lattice ; and @xmath26=0.54 , @xmath28 for the kagome lattice respectively.,width=336 ] in order to test for magnetic field coupling between the 1d half - fluxon chains , double facetted junctions ( fig .
[ fig : hilgfacets]c ) were also fabricated and tested .
it appeared that also in this case the phase coupling was stronger than the field coupling : fig .
[ fig : hilgquad]d shows a section of a 40@xmath12 m facet double junction that shows in - phase alignment between the two anti - ferromagnetically ordered 1d chains . because this arrangement places the positive half - fluxons in the lower left chain closest to the negative half - fluxons in the upper right chain , this is the lowest energy arrangement .
however , in sections of the 1d chains which show defects , as in the center of the upper right chains in fig .
[ fig : hilgquad]e and fig .
[ fig : hilgquad]f , the interchain alignment goes from in - phase to out - of - phase when the interchain ordering has a defect , but one chain does not develop a second defect to align the interchain spins .
therefore it appears that the energy cost to create a defect is larger than the energy gain from making neighboring chains in - phase .
, cooled in nominally zero field through the nb superconducting transition temperature at different cooling rates .
each panel is labelled with the cooling rate and experimentally determined bond order @xmath31 .
the white circles superimposed on the images label 6-ring loops in the honeycomb lattice in which the rings are perfectly anti - ferromagnetically ordered.,width=336 ] when the facetted junctions are cooled in an externally applied magnetic field , one spin direction becomes energetically favored over the other , but there is a competition between this energy and the anti - ferromagnetic coupling energy during the cooling process . an example is shown in figure [ fig : hilgfacs ] .
more detailed results and modelling of this facetted junction will be described in the next section .
c from fig .
[ fig : ec1vr]b , to determine which rings flipped sign after successive cooldowns.,width=192 ] in the electrically disconnected 2d lattices of fig .
[ fig : ec1sems ] , the square and honeycomb arrays are geometrically unfrustrated , as their magnetic moments can be arranged so that all nearest neighbors have opposite spins .
in contrast , the triangle and kagom lattices are geometrically frustrated , since it is impossible for all of the rings to have all nearest neighbors anti - ferromagnetically aligned .
[ fig : ec1imgs ] shows examples of scanning squid microscope images of the arrays of fig .
[ fig : ec1sems ] after cooling in nominally zero field .
although regions of perfect anti - ferromagnetic ordering are seen in the unfrustrated arrays ( fig .
[ fig : ec1imgs]a , b ) , anti - ferromagnetic ordering beyond a few lattice distances was never observed .
nevertheless , antiferromagnetic correlations were seen in all the 2d @xmath3-ring arrays .
a measure of the short range antiferromagnetic correlations is the bond order @xmath32 where @xmath33 is the fraction of the nearest neighbor pairs with opposite supercurrent circulation , and @xmath26(@xmath34 ) is the fraction of rings which have up ( down ) moments .
perfect antiferromagnetic correlation would correspond to @xmath31 = -1 .
the minimum possible bond order at zero applied field for the frustrated triangular and kagom lattices is @xmath31 = -1/3.@xcite the images in fig .
[ fig : ec1imgs ] are labelled with values for the bond orders .
it is to be expected that the anti - ferromagnetic ordering of the 2d arrays should improve if they are cooled more slowly through the nb superconducting transition .
figure [ fig : ec1vr ] shows squid microscope images of the same region of the honeycomb 2d lattice of fig .
[ fig : ec1sems ] , after cooling at various rates .
the individual panels are labelled with the cooling rates and final state bond orders .
the anti - ferromagnetic ordering increases with slower cooling rates .
one question that can be asked is : do particular regions of the 2d array order more strongly than others ? the white circles in fig .
[ fig : ec1vr ] outline the 6-member rings in the honeycomb arrays in which all neighbors are anti - ferromagnetically aligned .
this provides a convenient way of visualizing regions of local order .
it appears that there are no correlations between the positions of the ordered 6-member rings from cooldown to cooldown , and we conclude that the ordered regions are randomly distributed in space . in the @xmath7-ring experiments of davidovic
@xcite repeated cooling resulted in a particular ring often being in the same final state ( spin - up or spin - down ) .
this is presumably the result of the rings having slightly different effective areas , and therefore cooling in slightly different effective fields , since @xmath7-rings must be cooled in finite fields for there to be degenerate states .
the individual rings in our 2d @xmath3-ring arrays appear to have random final states .
figure [ fig : ec1flip ] shows a difference image between fig . [ fig : ec1vr]b and fig .
[ fig : ec1vr]c , which were taken after successive cooldowns . in this image ,
the rings which did not switch sign from one cooldown to the next are not visible , the rings which switched from down to up appear black , while the rings which switched from up to down appear white . of the roughly 663 rings in this field of view , 152 switched from down to up , and 162 switched from up to down .
this is consistent with random switching during cooldown . .
the junction has 10 faces , each 40 microns long , with alternating @xmath7- and @xmath3-intrinsic phase shifts.,width=336 ]
for the current purposes we treat the facetted ramp edge junction as a linear junction with alternating regions of @xmath7- and @xmath3- intrinsic phase shifts @xmath35 extending in the @xmath36-direction , with the junction normal in the @xmath37 direction , and the junction width @xmath38 in the @xmath39 direction small compared with the josephson penetration depth @xmath40 , where @xmath10 is the spacing between the superconducting faces making up the junction , and @xmath41 is the josephson critical current per area of the junction .
the quantum mechanical phase drop @xmath42 across the junction is the solution of the sine - gordon equation : @xmath43 analytical solutions to this equation are available,@xcite but here we solve eq .
[ eq : sine - gordon ] numerically.@xcite defining a dimensionless coupling parameter @xmath44 the differential equation eq .
[ eq : sine - gordon ] turns into a difference equation on a grid of size @xmath45 : @xmath46 taking the net current across the junction equal to zero , using a reduced externally applied magnetic field @xmath47 , and using the relation between the gradient of the phase and the field @xmath48 in the junction : @xmath49 we find boundary conditions that are described as difference equations , where @xmath50 is the total number of junctions : @xmath51 these coupled difference equations are solved using a relaxation method to find the solution @xmath42 , iterating to convergence . as the junction cools through the superconducting transition temperature t@xmath6 , the supercurrent density @xmath41 increases from zero , so that @xmath52 decreases from infinity . to model the cooling process
, we first solve eq .
[ eq : numerical_sg ] for @xmath52 much larger than the facet length @xmath53 , then decrease @xmath52 by a small amount , take the previous solution as the starting point for the next solution , and repeat until @xmath54 .
an example is shown in fig .
[ fig : oshilg ] . here
the externally applied flux was set at @xmath55 , the total junction length @xmath56=400@xmath12 m , @xmath45=1@xmath12 m , with 10 facets each of length @xmath53=40@xmath12 m , an initial @xmath57 , which was decreased to 4@xmath12 m in 24 equal steps .
the field threading the junction is given by @xmath58 .
the final solution strongly favors anti - ferromagnetic arrangement of the half - fluxon spins " . in several externally applied magnetic fields .
half - fluxons with magnetic fields in one direction are represented by open symbols , those with the opposite sense with closed symbols .
the left two panels are experimental results from two successive runs ; the right panel represents modelling as described in the text.,width=336 ] we can understand why the @xmath24 facetted junctions cool into perfect anti - ferromagnetic order , while the isolated linear arrays of @xmath3-rings do not , by considering the energetics of the cooling process . written as a difference equation ,
the free energy for a particular state of the facetted junction becomes : @xmath59 numerical solution of the sine - gordon equation as described above for the facetted junction of fig .
[ fig : hilgfacs ] shows that the free energy / facet , when the junction is in perfect anti - ferromagnetic ordering , is @xmath21 -5@xmath6010@xmath61k/@xmath52(@xmath12 m ) .
the energy cost to form a defect , by flipping one spin , is @xmath21 7.2@xmath6010@xmath61e@xmath62 .
this implies that when the free energy / facet is comparable to k@xmath63 t , the energy cost to form a defect is @xmath647.2@xmath6010@xmath61k : it is energetically favorable to form perfect anti - ferromagnetic ordering in linear @xmath24 facetted junctions cooled in zero field .
figure [ fig : hilgspi2 ] compares the results from repeated cooling of the 10-facet zigzag junction under various magnetic fields with modelling using the numerical solution of the sine - gordon equation outlined above .
note that there are some disorder effects in the cooling process , as evidenced by the slight differences between the two experimental runs .
the qualitative features of the data are reproduced by the modelling , although the experimental results are not as symmetric with respect to inversion in position , or with respect to field reversal , as predicted .
one possible source of the observed asymmetry might be field gradients .
however , putting a linear field gradient into the model did not improve the fit with experiment .
another source of asymmetry might come from the asymmetry of the junction and lead geometry . as a function of the reduced temperature .
@xmath65 is the barrier to thermally activated flipping of the sign of the spontaneous magnetization , @xmath66 is the strength of the spin - spin coupling energy , and @xmath67 is the thermal energy.,width=336 ] the cooling process for isolated @xmath3-rings is different than for electrically connected @xmath3-rings .
for simplicity we model our isolated @xmath3-rings considering a single inductance @xmath56 and josephson junction critical current @xmath68 .
the details of the formulas will be different for a two - junction ring,@xcite but we do not expect the physics to be qualitatively different .
the free energy u of such a @xmath3-ring can be written as @xmath69 where @xmath70 is the magnetic flux included in the ring , and @xmath71 is the flux externally applied to the ring . at temperatures important in the cooling process , close to the nb superconducting transition temperature t@xmath6 , field
screening effects can be neglected , since the magnetic penetration lengths are larger than the size of the sample . under these conditions
, we estimate the ring self - inductances ( for 2.7@xmath12 m junction width rings spaced by 11.5@xmath12 m ) to be @xmath643.9ph , and the mutual inductance m between nearest neighbor rings to be @xmath640.025ph . for a single junction @xmath3-ring spontaneous supercurrents flow when @xmath72 is just greater than 1 . the ring ordering process will occur in the regime @xmath73 . in this limit , and in the limit @xmath74 , where @xmath75 is the magnetic flux applied per ring
, the potential barrier to flipping the sign of the ring supercurrents in the @xmath76 ring can be approximated by @xmath77 with @xmath78 , where @xmath79 is the spin of the @xmath76 ring , and @xmath80 where @xmath71 is the externally applied flux and @xmath81 is the mutual inductance between the @xmath76 and @xmath82 rings . in the same limits ,
the value of the spontaneous flux @xmath83 generated by the @xmath76 ring is given by @xmath84 we take the junction critical current i@xmath6(0)=0.55ma for these rings , and the self - inductance of the rings is assumed to be temperature independent .
a measure of the spin - spin coupling energy @xmath85 is the difference between the energy barrier for all nearest neighbor spins up , minus that energy barrier if one of the neighbor spins is down .
[ fig : hilgengs ] plots the calculated barrier to thermally activated flipping @xmath65 , nearest neighbor coupling energy @xmath85 , and the thermal energy @xmath67 as a function of temperature for the honeycomb ring array of fig . [ fig : ec1imgs ] .
both e@xmath86 and e@xmath6 increase as the temperature is lowered until the flipping freezes out at a temperature @xmath87 such that @xmath88 , where @xmath89.@xcite for the rings of fig .
[ fig : ec1imgs ] , we estimate this occurs at @xmath90 = 3@xmath6010@xmath91 , with @xmath92 and @xmath93 . at this temperature @xmath94 . therefore antiferromagnetic coupling
is energetically favored in isolated @xmath3-rings , but not nearly as strongly as in facetted junctions .
calculated using monte - carlo techniques as described in the text , for various cooling rates @xmath95.,width=336 ] as the temperature is lowered through @xmath96 , the junction critical currents increase until two distinct circulating states become allowed for @xmath97 .
thermally activated switching between these states has a transition rate @xmath98 where @xmath99 is the junction hysteresis parameter , and @xmath100 is the junction characteristic time.@xcite we take the junction capacitance @xmath101 , @xmath102 , @xmath103 , and @xmath104=1@xmath105 for the 2d arrays presented in this paper .
we have taken the limits where the d.c .
shielding currents are much smaller than the critical currents , the spin - flip transition times are much shorter than the time between spin - flips , and @xmath106 .
we can isolate the temperature dependence of the prefactor of eq .
[ eq : rate ] by writing @xmath107 .
it is of interest to note that the power dissipated per ring near the freezout " temperature is of order @xmath108 .
to model the cooling process in our 2d ring arrays we use a metropolis monte carlo@xcite simulation : a 30x30 element array is set up with the same geometry as the experimental arrays .
boundary effects are minimized by using periodic boundary conditions : rings at one edge of the array are treated as if their nearest neighbors are the corresponding rings at the opposite edge .
the results of our simulations do not depend sensitively on whether periodic or non - periodic boundary conditions are used .
for each iteration cycle , the probability of a spin flip @xmath109 , @xmath110 is calculated for each ring . a random number between @xmath7 and @xmath111
is generated .
if this number is less than @xmath109 , the spin of the ring is flipped .
the process is repeated throughout the array , and the temperature is gradually reduced until no more spin flips occur .
figure [ fig : hilgsvt ] shows results from a monte - carlo simulation of the honeycomb lattice of fig . [ fig : ec1sems ] , using the ring parameters of table i , plotting the bond order @xmath31 as a function of the reduced temperature @xmath112 , for various values of @xmath95 , the change in @xmath112 per iteration cycle .
note that in this case the maximum temperature @xmath113 for which spontaneous magnetization is observed is @xmath114 . as the temperature is reduced
the spin flipping probabilities decrease , and anti - ferromagnetic ordering gradually occurs .
furthermore note that the freezing temperature is well below the temperature at which @xmath115 .
the effective cooling rate of the simulation is given by @xmath116 k - sec@xmath117 .
therefore the slowest cooling rate simulated is about @xmath118k / sec , much faster than the experimental cooling rates .
although it would in principle be possible to use sufficient computer time to match the modelled cooling rates to experiment , the modelling indicates stronger anti - ferromagnetic ordering than is observed , even at these very fast cooling rates . , and
as calculated for the same lattices using monte - carlo techniques as described in the text ( open symbols ) , for two values of the mutual inductance between nearest neighbor rings .
the calculations labelled 1st include only nearest neighbor spin - spin coupling , those labelled 1st+2nd also include second nearest neighbors , and those labelled 1st+2nd+3rd also include third nearest neighbors.,width=336 ] figure [ fig : hilgsvr ] shows experimental results for the final bond order for the honeycomb array of fig .
[ fig : ec1imgs ] , for various cooling rates .
experimentally the effect of cooling rate on final bond - order is weak .
also shown in this figure are the results of our monte - carlo simulations for the honeycomb array .
the points labelled @xmath119 use all of the parameters in the simulation as calculated following table i. the points labelled @xmath120 used a reduced value of the ring - ring mutual inductance . also shown are curves comparing the predictions if only the nearest neighbor spin - spin interactions are included , with results if 2nd and 3rd nearest neighbors are also included .
the mutual inductance between a ring and its further neighbors ( e.g. 2nd nearest and 3rd nearest ) is taken to be scaled by the cube of the ratio of the relevant center - to - center distances .
qualitatively , the inclusion of next nearest neighbors reduces the tendency towards anti - ferromagnetic ordering .
however , it appears that the inclusion of 3rd nearest neighbors does not change the results much further .
similarly , reducing the strength of the spin - spin coupling by a factor of 4 also reduces the tendency to order
. however , in all cases it appears that if the simulation were to be extended to sufficiently slow cooling rates to match experiment , long - range ordering would result . upon fraction of spin - up rings
@xmath26 , for the four different types of arrays shown in fig .
[ fig : ec1imgs ] , obtained by cooling in various fields .
the solid line is the theoretical maximum negative bond order for a geometrically unfrustrated array ; the dashed curve is that for a frustrated array .
some of the experimental points at very low and high @xmath26 fall below the theoretical maximum negative bond order because of edge effects due to the finite array sizes imaged .
( b ) results of a monte - carlo simulation of the cooling process for these arrays , as described in the text .
, width=336 ] this failure of our arrays to order implies that there is a source of disorder that we havent taken into account .
there are three additional sources of disorder that we have considered .
the first is a distribution in nb critical temperatures , and the second is a distribution in junction critical currents . in each case , our simulations indicate that an unrealistically large distribution width is required to significantly effect the predicted final @xmath31 value vs. cooling rate curves . in all the cases we have explored numerically : reducing @xmath65 , reducing @xmath85 , increasing the width of the t@xmath6 distribution , or increasing the width of the @xmath68 distribution , the @xmath31 vs. cooling rate curves can be shifted vertically in fig .
[ fig : hilgsvr ] , but the slope of these plots does not change qualitatively .
a final factor that might inhibit ordering of our arrays is if the cooling rate is not uniform .
however , inspection of fig .
[ fig : hilgsvt ] indicates that fast jumps in temperature of order 0.01@xmath121 would be required to significantly affect the final bond - orders .
it seems unlikely that our temperature sweeps are that non - uniform .
it is therefore difficult to see how the experimental data can be fit using the present simulations , and some factor that we havent considered correctly is causing the failure of our arrays to achieve long range order .
nevertheless , the experimentally determined values for the bond - order @xmath31 can be modelled fairly well , if one uses the cooling rate as a fitting parameter .
the left - hand panel of fig .
[ fig : hilgrel2]a summarizes cooling experiments for all of the 2d arrays of fig . [ fig : ec1imgs ] .
these data were taken after cooling the arrays at rates between 1mk / sec and 10mk / sec .
although long - range anti - ferromagnetic order was not observed in these arrays , they did show strong anti - ferromagnetic correlations , as evidenced by the large negative bond orders .
note that there are no qualitative differences between the geometrically frustrated and non - frustrated lattices in their tendency to anti - ferromagnetically order . for comparison , the @xmath7-rings of ref .
never showed bond - orders more negative than @xmath122 , whereas our @xmath3-ring arrays attained bond orders as negative as @xmath123 .
the right hand panel shows the results of monte - carlo simulations for these arrays , using the ring parameters in table i , except for @xmath120 instead of @xmath119 , and with a cooling rate @xmath124 . in both panels the ideal curve for an unfrustrated lattice ( solid line ) and a frustrated lattice ( dashed line )
are also indicated .
the experimental results can be qualitatively modelled assuming a spin - spin coupling constant about 4 times smaller than calculated , and a cooling rate @xmath125 times faster than the experiment .
we have shown that it is possible to use large arrays of photolithographically patterned @xmath3-rings as a model spin system .
half - fluxon josephson vortices in electrically connected 1d arrays , in the form of zigzag @xmath24 facetted junctions , order strongly anti - ferromagnetically through the superconducting order parameter phase .
electrically isolated 1d and 2d arrays order much less strongly .
the 2d @xmath3-ring arrays show stronger anti - ferromagnetic correlations than reported previously for @xmath7-ring arrays , but not long - range order beyond a few lattice constants .
we can understand the anti - ferromagnetic phase coupling of the 1d zigzag junctions by solving the sine - gordon equations .
although some features of the ordering in the 2d arrays can be understood using monte - carlo simulations , unrealistic parameters must be used , and we do not understand why these arrays do not show long range order .
we would like to thank d.h.a .
blank , r.h .
koch , k.a .
moler , g. rijnders , and h. rogalla for useful discussions .
we also acknowledge the dutch foundations fom and nwo for supporting this research .
deaver and w.m .
fairbank , phys .
lett . 7 , 43 ( 1961 ) .
r. doll and m. nbauer , phys .
lett . 7 , 51 ( 1961 ) .
d. davidovic et al .
76 , 815 ( 1996 ) .
d. davidovic et al .
b 55 , 6518 ( 1997 ) . l.n .
bulaevskii , v.v .
kuzii , and a.a .
sobyanin , jetp lett .
25 , 290 ( 1977 ) .
v.b . geshkenbein and a.i .
larkin , jetp lett .
43 , 395 ( 1986 ) .
geshkenbein , a.i .
larkin , and a. barone , phys .
b 36 , 235 ( 1987 ) .
wollman , d.j .
van harlingen , w.c .
lee , d.m .
ginsberg , and a.j .
leggett , phys .
71 , 2134 ( 1993 ) .
d.a . brawner and h.r .
ott , phys .
b 50 , r6530 ( 1994 ) . c.c .tsuei
et al . , phys .
73 , 593 ( 1994 ) .
a. mathai , y. gim , r.c .
black , a. amar , and f.c .
wellstood , phys .
74 , 4523 ( 1994 ) .
smilde et al . , phys .
88 , 057004 ( 2002 ) .
smilde , h. hilgenkamp , g. rijnders , h. rogalla , and d.h.a .
blank , appl .
80 , 4579 ( 2002 ) .
andreev , a.i .
buzdin , and r.m .
osgood , iii , phys .
b 43 , 10124 ( 1991 ) .
ryazanov , v.a .
oboznov , a.v .
veretennikov , and a. yu .
rusanov , phys .
b * 65 * , 020501(r ) ( 2001 ) .
a. bauer et al . , phys .
92 , 217001 ( 2004 ) .
h. hilgenkamp et al . ,
nature 422 , 50 ( 2003 ) .
rogers , master s thesis , mit , boston , ( 1983 ) .
l.n . vu and d.j .
van harlingen , ieee trans .
supercond . 3 , 1918 ( 1993 ) .
black et al . , appl .
62 , 2128 ( 1993 ) .
kirtley et al . , appl .
66 , 1138 ( 1995 ) .
d. xu , s.k .
yip , and j.a . sauls , phys .
b * 51 * , 16233 ( 1995 ) .
h. susanto , s.a .
van gils , t.p.p .
visser , ariando , h.j.h .
smilde , h. hilgenkamp , phys .
b * 68 * , 104501 ( 2003 ) .
kogan , j.r .
clem , and j.r .
kirtley , phys .
b 61 , 9122 ( 2000 ) .
kirtley , k.a .
moler , and d.j .
scalapino , phys .
b * 56 * , 886 ( 1997 ) .
e. goldobin , d. koelle , and r. kleiner , phys .
b 66 , 100508(r ) ( 2002 ) .
e. goldobin , d. koelle , and r. kleiner , phys .
b 67 , 224515 ( 2003 ) .
a. zenchuk and e. goldobin , phys . rev .
b 69 , 024515 ( 2004 ) .
e. goldobin , n. stefanakis , d. koelle , and r. kleiner , phys . rev .
b 70 , 094520 ( 2004 ) .
smilde , ariando , d.h.a .
blank , h. hilgenkamp , and h. rogalla , phys . rev .
b 70 , 024519 ( 2004 ) .
kirtley , c.c .
tsuei , and f. tafuri , phys .
90 , 257001 ( 2003 ) .
tesche , j. low . temp .
* 44 * , 119 ( 1981 ) . n. metropolis , a. w. rosenbluth , m. n. rosenbluth , a. h. teller , and e. teller , j. chem .
21 , 108 ( 1953 ) . | we report experiments in which one dimensional ( 1d ) and two dimensional ( 2d ) arrays of yba@xmath0cu@xmath1o@xmath2-nb @xmath3-rings are cooled through the superconducting transition temperature of the nb in various magnetic fields .
these @xmath3-rings have degenerate ground states with either clockwise or counter - clockwise spontaneous circulating supercurrents .
the final flux state of each ring in the arrays was determined using scanning squid microscopy . in the 1d arrays , fabricated as a single junction with facets alternating between alignment parallel to a [ 100 ] axis of the ybco and rotated @xmath4 to that axis , half - fluxon josephson vortices order strongly into an arrangement with alternating signs of their magnetic flux .
we demonstrate that this ordering is driven by phase coupling and model the cooling process with a numerical solution of the sine - gordon equation .
the 2d ring arrays couple to each other through the magnetic flux generated by the spontaneous supercurrents . using @xmath3-rings for the 2d flux coupling experiments eliminates one source of disorder seen in similar experiments using conventional superconducting rings , since @xmath3-rings have doubly degenerate ground states in the absence of an applied field . although anti - ferromagnetic ordering occurs , with larger negative bond orders than previously reported for arrays of conventional rings , long - range order is never observed , even in geometries without geometric frustration
this may be due to dynamical effects .
monte - carlo simulations of the 2d array cooling process are presented and compared with experiment . |
the importance of orbital degrees of freedom in understanding the complex phase diagrams of the manganites @xcite is subject of intense research activities ( see e.g. references @xcite for an overview ) .
the antiferromagnetic insulator lamno@xmath2 ( @xmath10 k ) is an orbitally ordered system @xcite , which has been established experimentally by resonant x - ray scattering @xcite and neutron diffraction @xcite
. moreover , saitoh _ _ et al .
_ _ recently reported evidence for orbital excitations by raman spectroscopy @xcite .
the orbital order in lamno@xmath2 is induced by the cooperative jahn - teller ( jt ) effect of the mn@xmath11 ions ( electronic configuration @xmath12 , spin @xmath13 ) , which at temperatures @xmath14 k leads to a strong orthorhombic distortion of the perovskite structure . in the paramagnetic state electron spin resonance ( esr )
reveals a single exchange - narrowed resonance line with a @xmath15 value near 2.0 due to all mn@xmath11 ions @xcite and hence directly probes the spin of interest . doping divalent ions like sr or ca onto the la@xmath11 place gradually suppresses the jt distortion and leads to a ferromagnetic insulating and finally to a metallic phase at approximately 15% sr doping . in a previous work , we presented a systematic esr study in single crystals of la@xmath16sr@xmath17mno@xmath2 with sr concentrations @xmath18 @xcite . in the jt distorted phase the resonance linewidth @xmath19
is strongly enhanced compared to the undistorted phase , reaching maximum values of about @xmath20 koe .
similar results were reported from polycrystalline la@xmath16ca@xmath17mno@xmath2 @xcite and oxygen doped ceramic lamno@xmath21 @xcite . moreover ,
the single crystals exhibit a pronounced anisotropy of the resonance linewidth in the jahn - teller distorted phase , which disappears at temperatures @xmath22 .
the pure lamno@xmath2 sample turned out to be strongly twinned and therefore did not allow a detailed analysis of the angular dependence .
instead , the la@xmath0sr@xmath1mno@xmath23 crystal was found to be untwinned and can effectively be treated like the mother compound ( @xmath24 ) , as it is still an antiferromagnetic insulator ( @xmath25 k ) showing a similar magnetic susceptibility . only the jt transition is shifted to lower temperatures , with @xmath26 k. in a first approach @xcite , we ascribed the orientation dependence of the esr linewidth to the influence of the dzyaloshinsky - moriya ( dm ) interaction , which arises from the tilting of the mno@xmath4 octahedra along the antiferromagnetically coupled @xmath27 axis , only .
however , the orthorhombic distortion of the mno@xmath4 octahedra itself gives rise to a crystal - field ( cf ) induced line broadening of comparable order of magnitude @xcite and for a complete description one has also to take into account the dm interaction via mn - o - mn bonds within the ferromagnetically coupled @xmath28 plane , which is smaller than along the @xmath27 axis but not negligible . in the present paper ,
we study the angular dependence of the resonance linewidth in the paramagnetic regime at temperatures @xmath29 .
the contributions of cf and dm interaction consist of the superposition of the four non equivalent mn ions in the orthorhombic unit cell .
the evaluation is based on the structural data for lamno@xmath2 determined from neutron scattering experiments by huang _
et al . _
the application to the experimental angular dependence of the esr linewidth at 200k and 300k allows to estimate the microscopic cf parameters @xmath5 and @xmath6 and the dm vectors for all mn - mn pairs . throughout this paper
we will use the crystallographic notation following huang _
et al . _ , where the @xmath27 axis ( instead of the former @xmath30 axis in reference @xcite ) denotes the direction perpendicular to the ferromagnetically coupled @xmath28 planes .
the esr measurements were performed with a bruker elexsys e500 cw - spectrometer at x - band frequency ( @xmath31 9.35 ghz ) , equipped with continuous gas - flow cryostats for he ( oxford instruments ) and n@xmath32 ( bruker ) , which allow to cover a temperature range between 4.2 k and 680 k. the esr spectra record the power @xmath33 absorbed by the sample from the transverse magnetic microwave field as a function of the static magnetic field @xmath34 .
the signal - to - noise ratio of the spectra is improved by detecting the derivative @xmath35 with lock - in technique .
a small single crystal of la@xmath0sr@xmath1mno@xmath2 ( volume 2 mm@xmath36 ) was fixed on a suprasil quartz rod with low - temperature glue ( general electrics ) , which allowed the rotation of the sample around defined crystallographic axes .
we measured the angular dependence of the paramagnetic resonance with respect to the orientation of the static magnetic field @xmath34 in the three crystallographic planes ( @xmath37 , @xmath28 , and @xmath38 ) at temperatures 200 k and 300 k , deeply in the jahn - teller distorted orthorhombic phase .
figure [ spectra ] illustrates typical esr spectra for the magnetic field applied parallel to the three main axes . in all cases
, one observes a broad , exchange - narrowed resonance line , which is well fitted by a dysonian line shape @xcite given by @xmath39 this is an asymmetric lorentzian line , which includes both absorption and dispersion , where @xmath40 denotes the dispersion - to - absorption ( d / a ) ratio . as the linewidth @xmath19 is of the same order of magnitude as the resonance field @xmath41 in the present compounds , eq .
( [ dyson ] ) takes into account both circular components of the exciting linearly polarized microwave field and therefore also includes the resonance at reversed magnetic field @xmath42 .
such asymmetric line shapes are usually observed in metals , where the skin effect drives electric and magnetic microwave components in the sample out of phase and therefore leads to an admixture of dispersion into the absorption spectra . for samples small
compared to the skin depth one expects a symmetric absorption spectrum ( @xmath40 = 0 ) , whereas for samples large compared to the skin depth absorption and dispersion are of equal strength yielding an asymmetric resonance line ( @xmath40 = 1 ) .
a second reason for the asymmetry , which also occurs in insulators , arises from the influence of non diagonal elements of the dynamic susceptibility : in systems with interactions of low symmetry and sufficiently broad resonance lines ( @xmath43 ) the line shape shows characteristic distortions depending on the frequency and orientation of the exciting microwave field @xcite , where eq .
( [ dyson ] ) yields a useful approximation of the spectrum . comparing the spectra at 200 k with those at 300 k
one recognizes an increasing asymmetry with increasing temperature : the average d / a ratio increases from about 0.05 at 200 k to 0.35 at 300 k. this can be understood in terms of the skin effect due to the increase of the conductivity with increasing temperature .
the resistivity @xmath44 of the sample under investigation is about @xmath45 @xmath46 cm at 200 k and @xmath47 @xmath46 cm at 300 k @xcite .
we estimated @xcite that for resistivities @xmath48 @xmath46 cm the skin depth becomes smaller than the dimensions of the sample , which are about 1 mm . as the skin depth is proportional to the square root of the resistivity , the influence of the skin effect becomes visible even at 300 k. however at 200 k , it can be neglected . the remaining asymmetry , which will be discussed in connection with the resonance field below , has to be ascribed to the non diagonal contributions of the dynamic susceptibility .
( upper frame ) , d / a ratio ( middle frame ) , and resonance field @xmath49 ( lower frame ) in la@xmath0sr@xmath1mno@xmath2 for the magnetic field applied within the three crystallographic planes at 200 k , width=302 ] figures [ exp200k ] and [ exp300k ] show the full angular dependence of linewidth ( upper frame ) , resonance field ( lower frame ) and d / a ratio ( middle frame ) for 200 k and 300 k , as obtained from the fit with eq .
[ dyson ] .
the linewidth exhibits a pronounced anisotropy with respect to the crystallographic @xmath27 axis and a weaker angular dependence within the @xmath28 plane , which can be empirically described by a @xmath50-law .
the d / a ratio depends on the orientation of the microwave field , which is applied parallel to the rotation axis , and even becomes negative at 200 k , if the microwave field is perpendicular to the @xmath28 plane .
this can not be understood in terms of the skin effect alone , which always produces a positive d / a ratio .
moreover , the influence of the skin effect should be negligible at 200 k. hence , the asymmetry has to be ascribed partly to the influence of the non diagonal elements of the dynamic susceptibility .
( upper frame ) , d / a ratio ( middle frame ) , and resonance field @xmath49 ( lower frame ) in la@xmath0sr@xmath1mno@xmath2 for the magnetic field applied within the three crystallographic planes at 300 k , width=302 ] the angular dependence of the resonance field exhibits additional @xmath51 modulations .
they appear to be strongest for the @xmath37 plane , where the d / a ratio exhibits the most pronounced angular dependence . at the same time the d / a ratio is strongly correlated with the resonance linewidth , as they both attain their maximum values at the same angle .
however , for these broad signals , the baseline is not so well defined as in the case of narrow lines and a shift in the d / a ratio can be compensated by resonance field and baseline . for this reason , the angular dependence of the d / a ratio can be an artificial effect , which is probably related to the large linewidth .
hence , we tried to fit the esr spectra with a constant average d / a ratio for each rotation plane .
it turned out that the spectra at 200 k can be satisfactorily fitted as well .
this procedure has no visible influence on the linewidth , but the resonance field is clearly changed , as can be obtained from fig .
[ r200kcor ] , which shows the corrected data for 200 k. the amplitude of the anisotropy remains approximately unchanged , but the additional @xmath51 modulations more or less disappear . turning to the resonance field data at 300 k , the situation is more complicated , as the influence of the skin effect becomes important , too .
the amplitude of the d / a - angular dependence increases and a fit with a constant d / a ratio is not satisfactory anymore .
therefore we omit a correction of the resonance fields at 300 k.
a strongly exchange coupled magnetic system like lamno@xmath2 can be described by the following hamiltonian @xmath52 where the first term describes the the superexchange interaction between two next - neighbor mn spins @xmath53 and @xmath54 with coupling constant @xmath55 .
the second term describes the zeeman splitting of the spin states with gyromagnetic tensor @xmath56 within an external magnetic field @xmath57 , where @xmath58 denotes the bohr magneton .
the third term @xmath59 includes all interactions , which do not conserve the total spin and therefore contribute to the broadening of the esr line as there are cf , dm interaction , anisotropic exchange ( ae ) interaction , dipole - dipole interaction , and hyper - fine interaction .
due to estimation of their relative strength @xcite it turned out that cf and dm interaction yield by far the largest contribution . to derive the appropriate expressions for cf and dm interaction , we use the structural parameters of the lamno@xmath2-iia sample ( space group _ pnma _ ) determined by huang _
et al . _
@xcite , because it shows an antiferromagnetic ground state and an ordering temperature of about 140 k , which is consistent with the magnetic properties of our sample @xcite .
figure 7(a ) in reference @xcite shows the crystallographic structure of lamno@xmath2 in the strongly jahn - teller distorted phase . due to the tilting of the mno@xmath4 octahedra
one can identify four inequivalent positions of manganese ions in the orthorhombic unit cell , which is illustrated in fig .
7(b ) of the same reference @xcite . in a local coordinate system , where the axes are directed along the mn - o bonds of the mno@xmath60 octahedra ,
the spin hamiltonian is usually written in cartesian spin components ( @xmath61 ) with parameters @xmath5 and @xmath6 as @xcite @xmath62 where the @xmath63 axis is directed along the longest mn - o link close to the @xmath28 plane and the @xmath64 axis is parallel to mn - o(1 ) bond almost along the @xmath27 direction .
small orthorhombic distortions of the mno@xmath60 octahedron will be neglected in this section .
because the linewidth is invariant under shift transformations , we can redefine @xmath65 for simplification as @xmath66 where @xmath67 and @xmath68 .
it is useful to recall that the @xmath64 axis is almost parallel to @xmath27 .
the octahedra are rotated with respect to their next neighbor within the unit cell .
therefore in the crystallographic system ( @xmath69 ) we have different crystal - field components at the four inequivalent mn places which are listed in table [ cfpara ] .
the cf components for the mn ion at @xmath70 read @xmath71 long , middle and short mn - o distances in square are @xmath72 here capital @xmath73 , @xmath74 and small letters @xmath75 , @xmath64 , @xmath63 denote the structure parameters ( after huang _ et al . _
@xcite ) for the oxygen ions along @xmath27 and within the ac plane respectively .
the hamiltonian , which describes the antisymmetric dm interaction @xcite , can be written as @xmath76,\ ] ] with the dm vector @xmath77 $ ] being perpendicular to the plane defined by a mn ion at site ( @xmath78 ) , the bridge ligand o , and a mn ion at site ( @xmath79 ) , where @xmath80 are unit vectors along mn@xmath11-o@xmath81 bonds @xcite .
the intrinsic scalar parameter @xmath82 strongly depends on the orbital states and the mn - o - mn bridge angle . in the case of
pure lamno@xmath2 both the tilting and the jt distortion of the mno@xmath4 octahedra account for the origin of antisymmetric contributions to the superexchange interaction between the mn ions .
a necessary condition for the existence of dm contributions is the lack of a center of inversion between the magnetic ions @xcite . with the apical oxygen being shifted away from the [ 010 ] axis
, there is a rather strong dm coupling between the @xmath28 planes and a smaller coupling within the @xmath28 planes .
figure [ dmchain ] depicts all next - neighbor couplings that give rise to a dm interaction . ,
@xmath83 , and @xmath84 denote the cartesian components of the dm vector @xmath85.,width=264 ] for the _ pnma _ structure , the components @xmath86 ( @xmath87 ) of the dm vectors of all mn pairs within the unit cell are listed in table [ dmpara ] .
the index @xmath88 denotes the four magnetically inequivalent positions of the mn ion in the unit cell .
the index @xmath89 refers to the six next neighbors around each mn site with number @xmath79 , as it is shown in fig .
[ dmchain ] .
the absolute values can be expressed via two parameters @xmath90 ( inter @xmath28 plane ) and @xmath91 ( intra @xmath28 plane ) as @xmath92 where @xmath93 measures the mn - o distance along the antiferromagnetically coupled @xmath27 direction and @xmath94 and @xmath95 denote the mn - o distances within the ferromagnetically coupled @xmath28-plane .
a microscopic expression for @xmath90 and @xmath91 is discussed in appendix 1 .
the resonance field of the strongly exchange narrowed esr line is generally determined by the first moment of the spectrum .
here we choose another approach , which also allows to take into account the demagnetization effect .
hence , we start from the equation of motion @xcite : @xmath96+ig\mu_{\rm b}m(n_{xx}-n_{yy})s^{-}\end{aligned}\ ] ] with the magnetization @xmath97 and the demagnetizing factors @xmath98 .
the circular spin operators are defined as usual @xmath99 .
here we assume an isotropic @xmath15 value , because the influence of the ae interaction can be neglected with respect to the crystal field .
the dm interaction does not contribute to the resonance shift , because the expectation value of the commutator @xmath100\rangle$ ] vanishes .
after linearization the expectation value of the commutator @xmath101\rangle$ ] is given by @xmath102\rangle & \approx & \frac{m_{\rm at}}{g\mu_{\rm b}}\frac{1}{4}\sum\limits_{j}(\tilde{d}_{xx}^{(j)}-\tilde{d}_{yy}^{(j ) } + 2i\tilde{d}_{xy}^{(j)})s^{- } \nonumber \\ & + & \frac{m_{\rm at}}{g\mu_{\rm b}}\frac{1}{4}\sum\limits_{j}(\tilde{d}_{xx}^{(j)}+\tilde{d}_{yy}^{(j ) } -2\tilde{d}_{zz}^{(j)})s^{+}\end{aligned}\ ] ] here @xmath103 $ ] is the magnetization per one mn site , and @xmath104 k is the paramagnetic curie - weiss temperature .
the crystal - field components @xmath105 refer to the coordinate system , in which the external field determines the @xmath63 axis and are related to the components @xmath106 in the crystallographic system via the usual transformation rules ( see appendix 2 ) .
the sum ( @xmath79 ) is running over the four inequivalent mn positions of the unit cell .
if we further neglect the demagnetization effect , which according to @xcite is relatively small in the paramagnetic regime , the esr frequency reads @xmath107^{2 } \\
\nonumber & -&{\frac{m_{\rm at}}{4g\mu_{\rm b}}}^2[\sum\limits_{j } ( \tilde{d}_{xx}^{(j)}-\tilde{d}_{yy}^{(j)}+2i\tilde{d}_{xy}^{(j ) } ) ] \\ \nonumber & \times & [ \sum\limits_{j } ( \tilde{d}_{xx}^{(j)}-\tilde{d}_{yy}^{(j)}-2i\tilde{d}_{xy}^{(j)})]\end{aligned}\ ] ] the angular dependence of the resonance field @xmath108 is obtained after substitution of @xmath105 by @xmath106 ( see appendix 2 ) .
polar angle @xmath109 and azimuth angle @xmath110 are measured with respect to the @xmath30 and @xmath111 axes of the orthorhobmic unit cell . in the case of strong exchange narrowing @xmath112 ,
the esr linewidth @xmath19 is determined by the second moment @xmath113 of the resonance line divided by the exchange frequency @xmath114 @xcite @xmath115 in the present case the second moment is determined by the sum of dm and cf contributions @xmath116 within the coordinate system , where the z axis is determined by the external magnetic field , the second moment due to dzyaloshinsky - moriya interaction is calculated as @xcite @xmath117}\end{aligned}\ ] ] the index @xmath88 is running over all four magnetically inequivalent positions of the mn ion in the unit cell .
the sum over index @xmath89 refers to the six next mn neighbors around each mn site with number @xmath79 .
after transformation into the crystallographic system and using the values listed in table [ dmpara ] , the average over all four positions yields the angular dependence @xmath118 \nonumber \\ & + & 4\beta _
{ ab}^{2}[1+\sin^{2}\theta\sin^{2}\varphi]\nonumber \\ & + & ( 2\gamma_{b}^{2}+4\gamma_{ab}^{2})[1+\cos^{2}\theta]\}\end{aligned}\ ] ] for the crystal field , the expression for the second moment is given in terms of @xmath119 by @xmath120^{2 } \nonumber \\ & + & 10f_{2}(\tilde{d}_{xz}^{2}+\tilde{d}_{yz}^{2})\nonumber \\ & + & f_{3}[(\tilde{d}_{xx } -\tilde{d}_{yy})^{2}+4\tilde{d}_{xy}^{2}]\}\end{aligned}\ ] ] and again the sum over @xmath79 is running over the four inequivalent mn positions in the unit cell . the prefactors @xmath121 allow to separate secular ( @xmath122 ) and non secular ( @xmath123 and @xmath124 ) contributions .
again , the angular dependence is obtained by transforming the cf components into the crystallographic system and using the values listed in table [ cfpara ] .
the full expressions divided in secular and nonsecular parts are given in appendix 2 .
it is important to point out here that both secular and nonsecular parts contain strong contributions proportional to @xmath125 and @xmath126 , which should result in a @xmath127 periodic modulation of the angular dependence .
however , in the sum @xmath128 these contributions cancel each other for the case @xmath129 = @xmath130 = @xmath131 .
as the experimental data do not exhibit any @xmath127 periodic modulation , only the latter case yields the appropriate description of the observed angular dependence . under these circumstances the crystal field contribution reads e.g. for the @xmath37 plane @xmath132\ { \frac{1}{2}\sum\limits_j [ 2d_{zz}^{(j)}-d_{xx}^{(j)}-d_{yy}^{(j)}]^{2 } \nonumber \\ & + & \frac{5}{2}\sum\limits_j[(d_{xx}^{(j)}-d_{yy}^{(j)})^{2}+4(d_{xy}^{(j)})^{2 } ] \nonumber
\\ & + & 7\sum\limits_j\left[(d_{zy}^{(j)})^{2}+(d_{xz}^{(j)})^{2}\right ] \\ & + & \sum\limits_j\left[3(d_{zy}^{(j)})^{2}-3(d_{xz}^{(j)})^{2}\right.\nonumber\\ & - & \left.(2d_{zz}^{(j)}-d_{xx}^{(j)}-d_{yy}^{(j ) } ) ( d_{xx}^{(j)}-d_{yy}^{(j)})\right ] \cos2\varphi \}\nonumber\end{aligned}\ ] ] the respective expressions for @xmath28 and @xmath38 plane are obtained by permutation of @xmath133 and exchange of @xmath134 , as indicated in appendix 2 .
as it is shown in fig .
[ linewidth ] , the linewidth data at 300 k are well described by the superposition of cf and dm contribution to the second moment ( eq . [ narrowed ] ) . in la@xmath0sr@xmath1mno@xmath2 for the magnetic field applied within the three crystallographic planes at 300 k.
the solid lines represent the fit with equation [ narrowed ] .
the lines below illustrate the contributions of cf and dm interaction , respectively , width=302 ] the superexchange integral @xmath55 was estimated in mean - field approximation from the curie - weiss temperature @xmath135 with ferromagnetic in - plane coupling @xmath136 to four neighboring mn ions and antiferromagnetic inter - plane coupling @xmath137 to two neighboring mn ions and the curie constant @xmath138 ( @xmath139 : boltzmann constant ) . assuming @xmath140 and inserting the mn@xmath11 spin @xmath141 and @xmath142 k @xcite
, we obtain @xmath143 k. both dm and cf interaction are of equal order of magnitude about 1k .
the dm interaction @xmath144 k along the antiferromagnetically coupled @xmath27 axis is about three times larger than @xmath145 k within the ferromagnetically coupled @xmath28 plane .
the absolute values of the crystal field parameters @xmath146 k and @xmath147 k are of nearly equal strength indicating comparable axial ( @xmath5 ) and rhombic ( @xmath6 ) distortions of the mno@xmath4 octahedra . the angular dependence of the resonance linewidth at 200 k
is well fitted with similar but slightly smaller parameters ( @xmath144 k , @xmath148 k , @xmath149 k , @xmath150 k ) in fig .
[ r200kcor ] .
in addition , the resonance field ( @xmath151 ) can be described by similar crystal field parameters as the linewidth , thereby showing good agreement of static and dynamic susceptibility .
( upper frame ) and corrected resonance field @xmath41 ( lower frame ) in la@xmath0sr@xmath1mno@xmath2 for the magnetic field applied within the three crystallographic planes at 200 k. the solid lines represent the fit with equations [ narrowed ] and [ resonance ] , respectively.,width=302 ] as the derivation of the linewidth was carried out in the high - temperature limit ( @xmath152 ) , where it approximates an asymptotic value @xmath153 , we have to take into account , that the fit parameters @xmath5 , @xmath6 , @xmath154 and @xmath155 should reflect the temperature dependence . according to huber _
et al . _
@xcite far apart from any magnetic or structural transitions , the temperature dependence can be approximated by the quotient of the single - ion curie susceptibility and the experimental curie - weiss susceptibility as @xmath156 taking into account that the square of cf and dm parameters appears in the formulae for the second moment , we can extrapolate these parameters to @xmath157 via the square root of eq .
[ huber ] . using the data at 300 k , which is far above magnetic order
, we have to multiply the parameters by a factor 1.26 and get @xmath158 k , @xmath159 k , @xmath160 k and @xmath161 k. comparing the values at 200 k , one expects a reduction of the parameters by approximately @xmath162 with respect to 300 k. indeed , this is well fulfilled , despite the onset of the critical behavior on approaching magnetic order at 140 k , only @xmath154 remains unchanged .
the equations for the resonance field already contain the temperature dependence within the magnetization @xmath163 .
for this reason it should be better described by the high - temperature values of the cf parameters than by the temperature dependent ones , which is the case at 200 k. due to the rather large uncertainty in the determination of the resonance field and the proximity of the critical temperature region , this discrepancy is not assumed to be of significant importance .
thus , at 300 k the amplitude of the uncorrected resonance - field anisotropy is nicely reproduced by the high - temperature values of the cf parameters as @xmath164 oe . the strength of cf and dm interaction was independently determined from orientation dependent magnetization data @xcite , measured at 4.2 k in a la@xmath0sr@xmath1mno@xmath2 single crystal of the same batch as used in our esr experiments . in combination with antiferromagnetic resonance measurements
it was shown there , that the ground state ( @xmath165 ) exhibits a canted antiferromagnetic structure , which rules out any phase separation in ferromagnetic and antiferromagnetic regions . within the two - sublattice model pimenov _
et al . _
obtained ( after translation to our hamiltonian in units of kelvin ) to the following parameters from the magnetization data @xcite : @xmath166 k , @xmath167 k , @xmath168 k , and @xmath169 k. these results are in good agreement with our findings , as the values of the cf parameters @xmath170 k and @xmath171 k are essentially equal .
the dm contribution is of similar strength , where @xmath155 was neglected .
thus , we could confirm the validity of their model from a microscopic point of view .
esr studies by huber _
et al . _ in lamno@xmath2 @xcite and tovar _ et al . _
@xcite in the series lamno@xmath21 revealed values comparable to our parameters . for pure lamno@xmath2 , they use the cf parameter @xmath172 k determined by moussa _
et al . _
from neutron - scattering experiments @xcite and estimate the dm contribution as @xmath173 k. for an oxygen excess @xmath174 with a transition temperature @xmath175 k comparable to la@xmath0sr@xmath1mno@xmath2 , the cf parameter @xmath5 is found to be about @xmath176 smaller than in the pure compound , whereas the dm interaction remains essentially unchanged at @xmath177 k. though these data were measured in polycrystalline samples , where no distinction between axial(@xmath5 ) and rhombic(@xmath6 ) cf parameters and the inter - plane and intra - plane dm interaction could be made , their results agree with ours within a factor of 2 .
the fact , that the two cf parameter are of comparable strength is of particular importance for the picture of orbital order in lamno@xmath2 .
many illustrations of the orbitally ordered state invoke the ordering of @xmath178 orbitals ( e.g. @xcite ) , as originally supposed by goodenough @xcite in order to describe the antiferromagnetic superexchange .
so far this picture has been widely applied and in the absence of an experimentum crucis , which directly probes not only the electron - electron interaction induced by orbital order ( which resonant x - ray scattering is assumed to be able to @xcite ) but also reveals the kind of orbitals involved ( which could not be identified by x - ray resonant scattering @xcite ) , it is the obvious choice in view of the observed axial distortion of the mno@xmath4 octahedra . considering the equal contributions of @xmath5 and @xmath6
, we conclude that this picture of an ordering of @xmath178 orbitals has to be modified as it results in @xmath179 in contradiction to our findings , which are also in agreement with the results of pimenov _ et al .
_ @xcite . in a local coordinate system
the orbital of the @xmath180 electron is a superposition @xmath181 @xcite , where @xmath182 and @xmath183 denote the orbital mixing coefficients down to lowest temperatures as reported by rodriguez - carvajal _ et al . _
@xcite on the base of neutron - diffraction studies .
matsumoto @xcite found in second order perturbation theory the following formula for the cf parameter @xmath6 @xmath184 with the spin - orbit coupling @xmath185 and the energy separation @xmath186 between the @xmath187 ground state and @xmath188 .
this formula qualitatively confirms our conclusions , as not only the sign of our parameters is reproduced but also both orbital mixing coefficients must not be negligibly small in order to explain the observed rhombic cf parameter e.
we performed a systematic investigation of the angular dependence of the paramagnetic resonance in the jahn - teller distorted orthorhombic phase in la@xmath0sr@xmath1mno@xmath2 single crystals .
we presented a comprehensive analysis for the resonance linewidth in high - temperature approximation , which takes into account the microscopic geometry of the four inequivalent mn positions in the orthorhombic unit cell based on the structural data determined for lamno@xmath2 by huang _
et al . _
the crystal - field parameters for all mn positions and the dzyaloshinsky - moriya interaction for nearest - neighbor mn ions along the @xmath27 axis as well as in the @xmath28 plane were successfully extracted as @xmath158 k , @xmath159 k , @xmath160 k and @xmath189 k. these findings shed new light on the microscopic picture of orbital ordering and spin - spin interaction in these compounds .
it is a pleasure to thank k .-
hck for fruitful discussions .
this work was supported in part by the bmbf under contract no .
13n6917 ( ekm ) , by the deutsche forschungsgemeinschaft ( dfg ) via the sonderforschungsbereich 484 and dfg - project
436-rus 113/566/0 , and by intas ( project no .
97 - 30850 ) .
the work of m. v. eremin was partially supported by rfbr grant no .
00 - 02 - 17597 and niokr tatarstan .
we consider the hamiltonian of superexchange coupling between two mn@xmath11 ions on lattice sites @xmath190 and @xmath191 @xmath192 where @xmath193 denote the superexchange parameters via one electron states , the symbol @xmath194 refers to the mn(@xmath195 ) state whereas @xmath196 refers to mn(@xmath197 ) . in the ground state mn@xmath11 ions
have maximum spin
. therefore we can write : @xmath198 along the @xmath27 axis , the superexchange is mainly realized via the @xmath199 channel and has antiferromagnetic character . within the @xmath28 plane the ferromagnetic coupling
is mainly transferred via the channel @xmath200 .
the relevant states like @xmath201 and @xmath202 are mutually orthogonal and yield negative @xmath203 .
the dzyaloshinsky - moriya parameter @xmath204 from equation [ dmham ] can be derived from the following expression @xmath205 here @xmath78 is the imaginary unit , @xmath206 and @xmath207 are spin - orbit coupling constants , @xmath208 denotes the energy splitting between the states @xmath196 and @xmath209 .
@xmath210 and @xmath211 are components of the one - electron orbital momentum perpendicular to the plane , built up by mn(@xmath195)-bridging oxygen(o ) -mn(@xmath197 ) , @xmath212 and @xmath213 are the corresponding unit vectors along the mn - o bonds .
@xmath214 and @xmath215 can be expressed via a product of dimensionless transfer integrals @xmath216 corresponding to mn - o bonds @xcite in order to explain the main features for the moment , we shall simplify the orbital ordering as @xmath202 like states along the @xmath27 axis and alternating ordering like @xmath217 , @xmath201 within the @xmath218 plane .
then one has the following situation : \1 ) along the @xmath27 axis , the largest contribution to the superexchange comes from @xmath219 , yielding @xmath220^{2}\ ] ] and for the dm vector the most important term is @xmath221 , which leads to @xmath222\lambda _ { \pi } \lambda _ { \sigma } \ ] ] \2 ) within the @xmath28 plane , the major contribution to the superexchange comes from @xmath223 and hence @xmath224^{2}\ ] ] whereas for @xmath225 vector the most important term is @xmath226 and therefore @xmath227\lambda _ { \pi } \lambda _ { \sigma } \ ] ] here @xmath228 is the angle between @xmath212 and @xmath229 vectors , @xmath230 denotes the intra - atomic hund s exchange parameter , @xmath231 is the charge - transfer energy between the mn ions . from these expression
one can clearly see that the relative sign of parameters @xmath204 along the @xmath27 axis ( @xmath90 in text ) and @xmath204 within @xmath218 plane ( @xmath91 in text ) are different as well as superexchange parameters @xmath232 and @xmath136 .
here we outline the transformation between the crystallographic system @xmath233 and the coordinate system @xmath234 with the @xmath235-axis parallel to the applied magnetic field @xmath34 and rotated by the polar angle @xmath109 and the azimuth angle @xmath110 with respect to the crystallographic system . assuming an isotropic @xmath15 value , the dm vector transforms like the space coordinates :
@xmath244\nonumber \\ & + & \sum\limits_{i , j}(g_{ij}^{y})^{2}[1+\sin^{2}\theta\sin^{2}\varphi]\\ & + & \sum\limits_{i , j}(g_{ij}^{z})^{2}[1+\cos^{2}\theta ] + \sum\limits_{i , j}g_{ij}^{x}g_{ij}^{y}\sin 2\varphi\sin^{2}\theta \nonumber \\ & + & \sum\limits_{i , j}g_{ij}^{x}g_{ij}^{z}\sin 2\theta \cos \varphi + \sum\limits_{i , j}g_{ij}^{y}g_{ij}^{z}\sin 2\theta \sin \varphi \}\nonumber\end{aligned}\ ] ] concerning the contribution of the crystal field @xmath249 , it is clear from values listed in table [ cfpara ] that many sums over four mn position vanish in the crystallographic system , which is very convenient .
for example @xmath250d_{yz}^{(j ) } \nonumber \\ & = & \sum\limits_j [ 2d_{yy}^{(j)}-d_{xx}^{(j)}-d_{zz}^{(j)}]d_{xz}^{(j ) } \nonumber \\ & = & \sum\limits_j [ 2d_{zz}^{(j)}-d_{xx}^{(j)}-d_{yy}^{(j)}]d_{xy}^{(j)}=0 \nonumber\end{aligned}\ ] ] from symmetry point of view it is understandable , because there are mirror planes within @xmath37 , @xmath28 and @xmath38 planes .
having in mind these properties , after some manipulation we arrive to the following formulae for the secular part of @xmath249 ( i.e. when factors @xmath130 and @xmath131 are equal to zero ) and nonsecular parts ( we consider the case @xmath130 = @xmath131 ) : @xmath251\{\frac{1}{4}\sum\limits_j [ 2d_{zz}^{(j)}-d_{xx}^{(j)}-d_{yy}^{(j)}]^{2 } \nonumber \\ & + & \frac{9}{8}\sum\limits_j[(d_{xx}^{(j)}-d_{yy}^{(j)})^{2}+4(d_{xy}^{(j)})^{2 } ] \nonumber \\ & -&\frac{3}{2}\sum\limits_j[2d_{zz}^{(j)}-d_{xx}^{(j)}-d_{yy}^{(j ) } ] ( d_{xx}^{(j)}-d_{yy}^{(j)})\cos2\varphi \nonumber \\ & + & \frac{9}{8}\sum\limits_j[(d_{xx}^{(j)}-d_{yy}^{(j)})^{2}-4(d_{xy}^{(j)})^{2 } ] \cos4\varphi \}\end{aligned}\ ] ] @xmath252\{\frac{1}{4}\sum\limits_j [ 2d_{zz}^{(j)}-d_{yy}^{(j)}-d_{xx}^{(j)}]^{2 } \nonumber \\ & + & \frac{11}{8}\sum\limits_j [ ( d_{yy}^{(j)}-d_{xx}^{(j)})^{2}+4(d_{xy}^{(j)})^{2 } ] \nonumber \\ & + & 7\sum\limits_j[(d_{zy}^{(j)})^{2}+(d_{xz}^{(j)})^{2 } ] \nonumber \\ & + & \sum\limits_j \{\frac{1}{2}[2d_{zz}^{(j)}-d_{yy}^{(j)}-d_{xx}^{(j ) } ] ( d_{xx}^{(j)}-d_{yy}^{(j)})\nonumber \\&+ & 3[(d_{zy}^{(j)})^{2}-(d_{xz}^{(j)})^{2}]\ } \cos2\varphi \nonumber \\ & -&\frac{9}{8}\sum\limits_j [ ( d_{xx}^{(j)}-d_{yy}^{(j)})^{2}-4(d_{xy}^{(j)})^{2}]\cos 4\varphi \}\end{aligned}\ ] ] a. k. bogush , v. i. pavlov , and l. v. balyko , cryst .
technol . * 18 * , 589 ( 1983 ) ; m. paraskevopoulos , f. mayr , c. hartinger , a. pimenov , j. hemberger , p. lunkenheimer , a. loidl , a. a. mukhin , v. yu .
ivanov , and a. m. balbashov , j. magn .
magn . mat .
* 211 * , 118 ( 2000 ) ; a. urushibara , y. moritomo , t. arima , a. asamitsu , g. kido , and y. tokura , phys . rev .
* b 51 * , 14103 ( 1995 ) ; j .- s .
zhou , j. b. goodenough , a. asamitsu , and y. tokura , phys .
* 79 * , 3234 ( 1997 ) .
j. p. hill , c. s. nelson , m. v. zimmermann , v .- j .
kim , doon gibbs , d. casa , b. keimer , y. murakami , c. venkataraman , t. gog , y. tomioka , y. tokura , v. kiryukhi , t.y .
koo , and s .- w .
cheong , cond - mat/01050643 . | we present a comprehensive analysis of dzyaloshinsky - moriya interaction and crystal - field parameters using the angular dependence of the paramagnetic resonance shift and linewidth in single crystals of la@xmath0sr@xmath1mno@xmath2 within the orthorhombic jahn - teller distorted phase .
the dzyaloshinsky - moriya interaction ( @xmath3 k ) results from the tilting of the mno@xmath4 octahedra against each other .
the crystal - field parameters @xmath5 and @xmath6 are found to be of comparable magnitude ( @xmath3 k ) with @xmath7 .
this indicates a strong mixing of the @xmath8 and @xmath9 states for the real orbital configuration . |
the effect of the heliospheric structure on gcrs propagation can be reproduced by a two dimensional ( radius and helio - colatitude ) stochastic model solving numerically the parkers s equation @xcite .
if we do not take into account the effects of the earth magnetosphere @xcite modulated fluxes depends not only on the level of solar activity but also on particles charge sign and solar magnetic field polarity @xcite .
the study of the modulation of @xmath0 ratios is particularly important , because it includes explicitly the combination of charge sign and polarity dependence .
the local interstellar spectra ( lis ) used as input of the code , both for protons and antiprotons , are taken by the galprop model@xcite .
the present code simulates the interactions of a gcr entering the heliosphere which extends up to a fixed distance of about 100 au from the sun .
one of the main parameters of the model is the diffusion tensor .
the parallel diffusion coefficient is : @xmath1=@xmath2 , where @xmath3 is the particle rigidity ( usually expressed in gv ) , @xmath4 is the diffusion parameter discussed in the next section and @xmath5 .
the perpendicular diffusion coefficient has two components , radial @xmath6 , and polar @xmath7 .
we used the relation : @xmath8 , where @xmath9 .
we also considered @xmath10 in the equatorial region , while we enhanced its value in the polar regions of the heliosphere @xcite : @xmath11 .
we used the tilt angle @xmath12 of the heliospheric current sheet ( hcs ) as a parameter for the level of the solar activity@xcite : the higher the value of @xmath12 the lower the expected gcr flux , for both solar field polarities .
values of the tilt angle are computed using two different models : the usual model uses a line - of - sight boundary condition at the photosphere and includes a significant polar field correction ; an alternative model uses a radial boundary condition at the photosphere , and requires no polar field correction . as suggested by ferreira and potgieter@xcite , the classical model is used for periods of increasing solar activity ( for example 20072012 , ams-01 data , ams-02 data ) , while the new model fits better for periods of decreasing solar activity ( for example 20002007 , bess data ) .
the three drift components do not depend on external parameters , except the solar polarity@xcite ( a@xmath130 for positive periods and a@xmath140 for negative periods@xcite ) .
the general drift expression is locally unlimited for a quasi - isotropic distribution@xcite , therefore we limit all the drift components below @xmath15 , which is the spatially averaged maximum value .
we selected gcr proton and antiproton data from 5 different experiments in order to compare and tune model results : ams-01@xcite , caprice@xcite , bess@xcite , heat and pamela @xcite .
the first two experiments took data in a period of positive solar polarity , bess in both solar polarities , and the last two in a@xmath140 period .
the corresponding periods of measurements are : june 1998 ( ams ) ; may 1998 ( caprice ) ; from july 1997 to december 2004 ( bess ) ; june 2000 ( heat ) ; and from 2007 to 2008 ( pamela ) . solar wind values for these periods
have been obtained from omniweb@xcite by 27 daily averages , while tilt angle values from the wilcox solar laboratory@xcite .
we can estimate the diffusion coefficient from a long term study@xcite of neutron monitor measurements and the _ _ force field model__@xcite ( ffm ) approach to the treatment of heliosphere transport of gcrs , using @xmath16 40 years of data ending in 2004 . - rising and declining phases for both negative and positive solar magnetic - field polarities - as a function of the ssn value ; the continuous lines are obtained from a fit of @xmath17 with respect to ssn values.,title="fig:",width=211 ] - rising and declining phases for both negative and positive solar magnetic - field polarities - as a function of the ssn value ; the continuous lines are obtained from a fit of @xmath17 with respect to ssn values.,title="fig:",width=211 ] - rising and declining phases for both negative and positive solar magnetic - field polarities - as a function of the ssn value ; the continuous lines are obtained from a fit of @xmath17 with respect to ssn values.,title="fig:",width=211 ] - rising and declining phases for both negative and positive solar magnetic - field polarities - as a function of the ssn value ; the continuous lines are obtained from a fit of @xmath17 with respect to ssn values.,title="fig:",width=211 ] in ffm@xcite [ e.g. , see also section 4.1.2.3 of leroy and rancoita ( 2009 ) ] , gleeson and axford ( 1968 ) assumed that i ) modulation effects can be expressed with a spherically symmetric modulated number density @xmath18 of gcrs - the so - called differential density - with kinetic energy between @xmath19 and @xmath20 , ii ) the diffusion coefficient at the time @xmath21 is given by a separable function of @xmath22 ( the radial distance from the sun ) and @xmath3 ( the particle rigidity in gv ) : @xmath23 with @xmath24 , @xmath25 the particle velocity , @xmath26 the speed of light , @xmath27 for particle rigidities above @xmath28gv and iii ) the modulation occurs in a steady - state condition , i.e. , the relaxation time of the distribution is short with respect to the solar cycle duration so that @xmath29 they derived that the differential intensity at a radial distance @xmath22 is given by the expression @xmath30,\ ] ] where @xmath31 is the undisturbed intensity beyond the solar wind termination located at a radial distance @xmath32 from the sun ; @xmath33 is the total energy of the particle with rest mass @xmath34 and , finally , @xmath35 is the so - called force - field energy loss@xcite .
when modulation is small@xcite - i.e. , @xmath36 - , they determined that @xmath37 where @xmath38 is the particle charge and @xmath39 is the so - called modulation strength ( or modulation parameter ) . assuming that @xmath40 ( the solar wind speed ) and @xmath4 are almost constant , @xmath39 - usually expressed in units of gv ( or mv ) - reduces to @xmath41 from which one gets @xmath42 i.e. , @xmath4 is linearly dependent on @xmath43 . in the ffm ,
the diffusion coefficient @xmath44 is scalar quantity and , as a consequence , does not account for effects related to the charge sign of the transported particles .
@xmath45 is independent of the species of gcr particles [ e.g. , see discussion at page 1014 of gleeson and axford ( 1968 ) or equation ( 1 ) of usoskin and collaborators ( 2005 ) ] .
the values of the modulation strengths [ @xmath46 were monthly determined for the time period@xcite from 1951 up to 2004 using measurements of neutron monitors ( i.e. , located at @xmath47au ) ; while those for the solar wind speeds are available on the web@xcite .
it has to be remarked that @xmath4 depends on the value of the solar wind termination located at a radial distance @xmath32 related , in turn , also to the solar wind speed [ e.g. , see chapter 7 of meyer - vernet(2007 ) and section 4.1.2.2 of leroy and rancoita ( 2009 ) ] .
however , because the present simulation code assumes a fixed solar wind termination at 100au to calculate the modulated differential intensities at @xmath48 , one has to derive from the diffusion parameter @xmath4 that ( @xmath49 ) for an effective heliosphere with a radial extension of 100au ( see sect . [
subsec : par ] ) .
thus , using eq .
( [ md_par_red2 ] ) one can obtain @xmath50,\ ] ] where 99au is the distance of the earth from the border of the effective heliosphere as defined in the current simulation code . in fig .
[ fig_sim1 ] , the diffusion parameter @xmath17 - obtained from eq .
( [ md_par_red1 ] ) - is shown as a function of the corresponding smooted sunspot number ( ssn ) value .
ratio as a function of the kinetic energy in gev at 1 au in comparison with experimental data : bess ( 1995 - 1997).,width=384 ] ratio as a function of the kinetic energy in gev at 1 au in comparison with experimental data : pamela ( 2007 - 2008).,width=384 ] the @xmath17 data had to be subdivided in four sets , i.e. , rising and declining phases for both negative and positive solar magnetic - field polarities .
for each set , the data i ) could be fitted with a relationship - indicated in in fi .
[ fig_sim1 ] - between @xmath17 and smoothed sunspot numbers ssn@xcite values and ii ) exhibited a gaussian distribution of difference of @xmath17 values from the corresponding fitted values .
the rmss of the gaussian distributions were found to be @xmath51 for the phases rising with @xmath52 , declining with @xmath52 , rising with @xmath53 , declining with @xmath53 , respectively . in this way - i.e. , from the relationships found - , we can use the estimated ssn values to obtain the diffusion parameter @xmath17 at times beyond 2004 . in practice , this procedure allows one to extend the @xmath54 years period by exploiting the linear relationship between the fitted @xmath49 values and the ssn values ( one of the main parameters related to the solar activity ) .
thus , we introduced in our code a gaussian random variation of @xmath17 with rmss corresponding to those found for each subset of data .
results of the simulation with and without the gaussian variation are consistent inside the uncertainties of the code @xcite .
our code simulates the interactions of a gcr entering the heliosphere from its outer limit , the helio - pause , located - as already mentioned - approximately at 100 au@xcite , and moving inwards to the earth located at 1 au .
we evaluated the time @xmath55 needed to sw to expand from the outer corona up to the helio - pause . considering an average speed of 400 km
/ s it takes nearly 14 months . while the time interval @xmath56 of the stochastic evolution of a quasi particle inside the heliosphere from 100 au down to 1 au is @xmath16 1 month at 200 mev and few days at 10 gev .
this scenario , where @xmath57 and @xmath58 1 month , indicates that we can not use fixed parameters ( monthly averages ) to describe the conditions of heliosphere in the modulation process .
in fact at 100 au , where particles are injected , the conditions of the solar activity are similar to the conditions present at the earth roughly 14 months before
. therefore we consider @xmath56 negligible with respect to @xmath55 and divide the heliosphere in 14 regions , as a function of the radius .
for each region we evaluated@xcite @xmath17 , @xmath12 and @xmath59 , in relation to the expansion velocity , in a dynamic way . in the future
the time spent by a gcr particle inside the heliosphere , as a function of the stochastic path and of the particle energy , will be also taken into account .
ratio as a function of the kinetic energy in gev at 1 au for ams-02 experiment ( january 2011).,width=336 ] we performed the simulations using dynamic values of @xmath17 , @xmath12 and @xmath59 .
results are shown in .
simulated fluxes with dynamic values show a very good agreement with measured data , within the quoted error bars .
this happens both in periods with a@xmath130 , in comparison with bess , and in periods with a@xmath140 , in comparison with pamela .
this means that our dynamic description of the heliosphere improves the understanding of the complex processes occurring inside the solar cavity .
the periodic behavior of the heliosphere allows us to predict , with a certain level of precision , the parameters needed for a simulation related to a time in the near future . in order to get these data we considered the prediction of ssn from ips ( ionospheric prediction service ) of the australian bureau of meteorology@xcite
we concentrate our simulations on the ams-02 mission@xcite that will be installed on the iss in february 2011 , and , in particular , at a time approaching the solar maximum : january 2012 .
we show in fig .
[ fig_sim4 ] the predictions of gcr modulation for the antiproton / proton ratio .
we built a 2d stochastic monte carlo code for particles propagation across the heliosphere .
present model takes into account drift effects and shows quantitatively a good agreement with measured values , both for positive and negative periods and for different particles and charge sign .
this is relevant because particles with opposite charge sign undergo a different solar modulation@xcite .
we compared our simulations with antiproton / proton ratios measured by bess and pamela .
we used _ dynamic _ parameters values ( @xmath17 , @xmath12 and @xmath59 ) for the related periods , in order to reproduce the propagation of incoming gcr through magnetic disturbances carried by the outgoing solar wind .
the _ dynamic _ description of the heliosphere and the forward approach seem to reproduce better the real physical propagation of gcr in the solar cavity . in order to have a more sophysticated model we need to introduce a dependence on the particle time spent in the heliosphere and a larger statistics of measured data during negative solar field periods , as ams-02 will provide in the next years .
recent measurements@xcite have pointed out the needs to reach a high level of accuracy in the modulation of the fluxes , in relation to the charge sign of the particles and the solar field polarity@xcite .
this aspect will be even more crucial in the next generation of experiments@xcite .
p. bobik , g. boella , m.j .
boschini , m. gervasi , d. grandi , k. kudela , s. pensotti , and p.g .
rancoita , proton modulation in the heliosphere for different solar conditions and prediction for ams-02 , proc . of the icatpp conference on cosmic rays for particle and astroparticle physics , october 78 2010 , villa olmo , como , italy , giani , s. , leroy , c. and rancoita , p.g . , editors , world scientific , singapore , isbn 978 - 981 - 4329 - 02 - 6 ; arxiv:1012.0938 [ astro-ph.ep ] , available at the web site : + _
http://arxiv.org / abs/1012.0938_. p. bobik , g. boella , m.j .
boschini , s. della torre , m. gervasi , d. grandi , k. kudela , s. pensotti , and p.g .
rancoita , `` drift models and polar field for cosmic rays propagatin in the heliosphere '' , proc . of the 11th icatpp , como 5 - 9/10/2009 , world scientific , singapore , 760 ( 2009 ) . | galactic cosmic rays ( gcrs ) are mainly protons confined in the galactic magnetic field to form an isotropic flux inside the galaxy . before reaching the earth orbit
they enter the heliosphere and undergo diffusion , convection , magnetic drift and adiabatic energy loss .
the result is a reduction of particles flux at low energy ( below 10 gev ) , called solar modulation .
we realized a quasi time - dependent 2d stochastic simulation of solar modulation that is able to reproduce cr spectra once known the local interstellar spectrum ( lis ) .
we were able to estimate the different behaviors associated to the polarity dependence of the heliospheric modulation for particles as well as for antiparticles .
we show a good agreement with the antiproton / proton ratio measured by ams-01 , pamela , bess , heat and caprice and we performed a prediction for the ams-02 experiment .
accepted for publication in the proceedings of the icatpp conference on cosmic rays for particle and astroparticle physics , + villa olmo ( como , italy ) , 78 october , 2010 , + to be published by world scientific ( singapore ) . |
during the last decade much attention has been devoted to theoretical and experimental investigations of the pionic double charge exchange ( @xmath17dcx ) process on nuclei .
because this reaction @xmath18 involves ( at least ) two nucleons in the nucleus , the @xmath17dcx cross section depends sensitively on short - range nn - correlations in nuclei .
therefore , it provides a good testing ground for the nucleon - nucleon interaction at short range .
experiments on different nuclear targets have unambiguously confirmed the existence of a narrow resonance - like structure in the @xmath17dcx cross - section at small incident pion energies @xmath19 mev @xcite .
the position of this peak turned out to be largely independent of the studied nucleus .
the height and width of this peak could not be explained by standard calculations based on the two - step process @xcite ( @xmath20 ) .
so far , these data could only be explained with the assumption of a non - nucleonic reaction mechanism @xcite proceeding via an intermediate dibaryon resonance , henceforth called @xmath1 .
the quantum numbers of the @xmath1 dibaryon candidate were determined as j@xmath2=0@xmath3 , t=0 , and its free mass and hadronic decay width were suggested to be @xmath10=2.065 gev and @xmath21 0.5 mev .
more than a decade ago mulders _
et al . _
@xcite predicted a dibaryon resonance with quantum numbers @xmath22=@xmath23 , t=0 and a mass @xmath24 mev within the mit bag model .
recently , this dibaryon candidate has been investigated in a series of works @xcite within the tbingen chiral constituent quark model .
these works emphasize the crucial role of the confinement mechanism for the existence of the @xmath1 .
the quantum numbers j@xmath2=0@xmath3 , t=0 of the @xmath1 resonance prevent the decay into two nucleons and the only allowed hadronic decay channel of the @xmath1 is the three - body decay into a @xmath25 system with s - waves in each particle - pair @xcite . because the @xmath1 mass @xmath26 is only @xmath27 mev above the @xmath25 threshold , the @xmath1 decay width @xmath28 should be anomalously small owing to a very small phase volume of three - particle final states .
we recall that the currently available experimental evidence of dibaryon excitations in nuclei is very limited @xcite .
this is due to very large n - n decay widths of most dibaryon resonances , which renders them undetectable on the background of other hadronic processes at intermediate energy . at present , the experimental evidence for narrow dibaryons is reduced to a single candidate , the @xmath29 .
in contrast to the deuteron , which consists of two on the average widely separated nucleons , there are indications @xcite , that the @xmath1 is a rather pure , compound six - quark system .
therefore , the dynamics of its hadronic decay into the @xmath17nn system should be sensitive to the overlap region of the two outgoing nucleons ; a situation that is ideal for understanding the role of quark degrees of freedom in the short - range nucleon - nucleon interaction ( see e.g. ref .
@xcite and references therein ) .
starting from this point ( for alternative approaches see in refs .
@xcite ) we consider the @xmath1 decay as a ( quark ) shell - model transition from one six - quark configuration to another one by emitting a pion .
the quark line diagram of the decay is sketched in figure [ figure : obufig1 ] .
the calculation of the transition matrix elements @xmath0 is similar to the calculation of @xmath30-isobar - decay matrix elements @xmath31 ( spin and isospin flip of a quark ) . in the case of the @xmath1 decay
only the initial dibaryon state is a definite six - quark configuration ( the lowest shell - model configuration with quantum numbers j@xmath2=0@xmath3 , t=0 ) , whereas the final state consists of a continuum of nn - states which have to be projected onto a basis of six - quark configurations with quantum numbers j@xmath2=0@xmath32 , t=1 of the nn @xmath33 wave .
the main difficulty in comparing the calculated width @xmath34 with experimental data is its sharp dependence on the energy gap between @xmath26 and the @xmath17nn threshold
. a reliable result on @xmath28 can be obtained only if the exact mass @xmath26 in vacuum is known ( e.g. from electroexcitation of the @xmath1 on the deuteron at large momentum transfers @xcite ) . at present , we have only indirect data in the nuclear medium @xcite . due to the absence of vacuum data ,
we investigate the problem of the @xmath1 decay width starting from theoretical quark - model results @xcite for @xmath26 and the hadronic @xmath1 size parameter @xmath35 .
our first calculation for @xmath28 was published in ref .
the aim of the present work is to improve mainly on three important effects which were neglected in ref .
@xcite : a ) antisymmetrization of the final nn - state on the quark level taking into account the effect of quark exchange between the two nucleons at short range , b ) insertion of a complete basis of final six - quark states including besides the non - excited @xmath36 shell - model state all pauli - allowed excited configurations @xmath37 and @xmath38 , which have a non - vanishing overlap with the final nn - state and can be populated via the emission of the pion from the initial @xmath1 dibaryon , and c ) inclusion of the final state interaction ( f.s.i . ) for the two - nucleon system .
as in ref . @xcite we consider only the simplest six - quark configuration @xmath5_x$ ] in the initial state ( the energetically lowest j@xmath2=0@xmath3 , t=0 translationally invariant shell model ( tism ) state which satisfies the pauli exclusion principle ) . it has been shown in @xcite , that the @xmath1 wave function may be considered as a compound six - quark state , for which a single shell model vector provides an adequate description .
this state vector is defined by @xmath39_s:[21 ^ 4]_{cts},\,lst\!=\!110,\,j^p\!=0 ^ -\rangle \ ; .\ ] ] the characteristic oscillator parameter in the six - quark wave function @xmath35 may for example be determined from the minimisation of the @xmath1 mass for a given microscopic quark - quark hamiltonian @xcite .
the young schemes @xmath40 $ ] , d = x , c , s , t in orbital , color , spin and isospin space , as well as for the coupled spaces ct , cts , are necessary for the unambigious classification of shell - model basis vectors in terms of irreducible representations ( ir ) of the following reduction chain for unitary groups : @xmath41 the fractional parentage coefficient ( f.p.c . )
technique @xcite based on scalar factors ( sf ) of clebsch - gordan coefficients of the above group @xcite , sketched in the following section , is used for the calculation of matrix elements and overlap integrals .
the pionic decay width of the @xmath1 is calculated , as in ref .
@xcite , assuming a direct coupling of constituent quarks with the isotriplet of pion fields @xmath42 through the operator @xmath43 here , @xmath44 , @xmath45 and @xmath46 are coordinate , spin and isospin of the j - th quark , * k * is the pion momentum in the c.m.s . of the @xmath1 , and @xmath47 .
@xmath48 is the @xmath49 coupling constant .
its value is connected with the @xmath25 coupling @xmath50 ( we use @xmath51 ) through the known relation @xmath52 , giving @xmath53 .
because we neglect isospin breaking effects in this work , we chose the average pion mass @xmath54 = 138 mev . in ref .
@xcite the wave function of the final nn - state was antisymmetrized and normalized on the nucleon level , assuming a plane wave with wave vector @xmath55 in the relative coordinate @xmath56 between the two nucleons .
the coordinate representation of the nucleon - nucleon state vector @xmath57 was then written as @xmath58 \ ; ; \ ; s=0 , t=1 \ , .\ ] ] the full wave function of the final state respected the three - quark cluster nature of the nucleons @xmath59 i.e. the nucleon wave function @xmath60 was given by translationally - invariant shell - model configurations @xmath61_{cts},lst\!=0,\ , 1/2,\ , 1/2\rangle_{tism } \nonumber \\ & = & \phi_n(123 ) \,\cdot \ , \chi^{c=0}\cdot\chi^{s=\frac{1}{2}}_{s_z}\cdot\chi^{t=\frac{1}{2}}_{t_z}\ ; .\end{aligned}\ ] ] @xmath62 , @xmath63 and @xmath64 are color - singlet , spin and isospin three - quark states .
@xmath65 is the orbital part of the wave function , expressed in terms of the internal jacobi coordinates @xmath66 and @xmath67 @xmath68 \ ; , \ ] ] with a characteristic nucleon oscillator parameter @xmath69 .
this parameter does not have to be the same as the harmonic oscillator parameter @xmath35 for the dibaryon wave function of eq .
( [ d1 ] ) , as it has been discussed in refs .
the relative jacobi coordinate between the clusters is given by @xmath70 note that the six - quark final state ( [ tpsinn ] ) is antisymmetrized automatically when the state vector @xmath71 is substituted into the decay matrix element @xmath72 ) is fully antisymmetric .
however , the antisymmetrizer - projector @xmath73 ( @xmath74 ) contained in the initial state @xmath75 reduces considerably the normalization of the final state ( it cuts all non - antisymmetrized parts of the cluster function of eq .
( [ tpsinn ] ) which contain about 90% of the wave function see below ) .
therefore , it is important to substitute from the beginning a final state wave function which is normalized ( @xmath76 ) and antisymmetrized ( @xmath73 ) on the quark level .
@xmath77 the normalization factor @xmath78 is determined by the standard orthonormalization condition @xmath79 which leads to @xmath80 the antisymmetrizer - projector @xmath73 is @xmath81 @xmath82 is the pair - permutation operator for quarks @xmath83 and @xmath84 in orbital , color , spin and isospin space .
it is instructive to calculate the normalization factor ( [ norm2 ] ) algebraically by factorization of the cst and x parts of the pair permutation @xmath85 .
the matrix element of @xmath86 between two nn - states in the st=01 ( or 10 ) channel is very well known ( see e.g. @xcite ) @xmath87 inserting this value into eq .
( [ norm2 ] ) reduces its right - hand - side to @xmath88 where @xmath65 is the orbital part of the nucleon wave function ( [ n ] ) given in eq .
( [ phin ] ) . the numerator in eq .
( [ normred ] ) depends on the form of @xmath89 , but for the plane wave in eq .
( [ pw ] ) ( or for any continuum wave function including f.s.i . )
it is a finite value , i.e. it has to be zero compared with the @xmath90-function in the denominator .
therefore , in our case the second term in eq .
( [ normred ] ) vanishes and we obtain @xmath91 note , that the expression @xmath92 receives its maximal value = 1 in the special case of a gaussian @xmath93 . therefore , for any relative nn wave function @xmath94 we have the following constraints : @xmath95 the value ( [ nrm ] ) is equal to the usual identity factor @xmath96 well - known in nuclear cluster physics ( see e.g. @xcite ) .
it plays an important role for the projection of six - quark configurations onto baryon - baryon channels @xcite . from now on we shall omit the antisymmetrizer @xmath73 in front of the final state in the decay matrix element , but the identity factor ( [ nrm ] ) may not be omitted : @xmath97 the inclusion of this factor , due to the antisymmetrization of the final two - nucleon wave function on the quark level , improves considerably the agreement of the results obtained in ref .
@xcite with the experimentally suggested width . as in ref .
@xcite , we calculate the decay matrix element of eq .
( [ rnm ] ) by inserting a complete set of six - quark configurations with quantum numbers of the final @xmath33 two - nucleon state ( lst=001 , j@xmath2=0@xmath32 ) @xmath98_x,\ , [ 2 ^ 21 ^ 2]_{ct}lst\!=\!110,\ , j^p\ ! = \ ! 0 ^ - \rangle \ ; . \label{tme}\end{aligned}\ ] ] here , @xmath99,\,[f_{ct}]\}$ ] and ( n ) defines quark states with @xmath100 harmonic oscillator ( h.o . ) excitation quanta , i.e.@xmath101 , _ etc . _ ; and @xmath35 is the h.o .
parameter for the six - quark system .
the summation in eq .
( [ tme ] ) extends over a limited set of young schemes @xmath102 $ ] and @xmath103 $ ] : the possible representations of @xmath103 $ ] in the sum in eq .
( [ tme ] ) are given by the series of inner products of the @xmath104_c$ ] color and t=1 @xmath105_t$ ] isospin young schemes @xmath106_c\circ[42]_t=[42]_{ct}+[321]_{ct}+[2 ^ 3]_{ct}+[31 ^ 3]_{ct}+[21 ^ 4]_{ct } \ ; .\ ] ] only two spatial young schemes @xmath107_x$ ] and @xmath105_x$ ] are compatible with the even - parity ( l=0 ) n - n partial wave .
further constraints follow from the pauli exclusion principle , i.e.@xmath102\circ [ 33]_s\circ [ f_{ct } ] = [ 1 ^ 6]_{xcst}$ ] . in the case of full spatial symmetry @xmath107_x$ ] , only one color - isospin state @xmath104_{ct}$ ] is allowed , but the young scheme @xmath105_x$ ] of the excited shell - model configurations is compatible with each state from the inner product given in eq .
( [ ser ] ) .
our choice of a one - body transition ( pion - production ) operator defined in eq .
( [ o ] ) further restricts the number of relevant intermediate states .
the one - particle operator ( [ o ] ) can excite ( or de - excite ) only one quark of the initial @xmath108 state .
therefore , the complete set of states in eq .
( [ tme ] ) is reduced to the configurations @xmath36 , @xmath37 and @xmath38 , knowing that higher one - particle excitations can be omitted because of a very small overlap with the final nn - state .
summarizing , the following intermediate states are taken into account in eq .
( [ tme ] ) : * the energetically lowest ( n=0 ) spatially symmetric state @xmath6_x[2 ^ 3]_{ct}$ ] , * the excited ( n=2 ) translationally - invariant ( orthogonalized to the 2s excitation of the six - quark c.m . ) state @xmath109 with identical young schemes @xmath107_x , [ 2 ^ 3]_{ct}$ ] , and * five excited ( n=2 ) states @xmath7_x[f_{ct}]$ ] with ct young schemes from the inner product of eq .
( [ ser ] ) .
it is interesting to note that all these configurations are also important for explaining the short - range nucleon - nucleon interaction .
this was pointed out almost two decades ago @xcite and thereafter discussed in many papers ( see e.g. @xcite and references therein ) .
now we believe that a possible @xmath1 dibaryon has much potential for providing additional information on the innermost part of the nucleon - nucleon interaction , i.e. in the region where the nucleons overlap . to take into account the f.s.i . for the two outgoing nucleons , we consider separable - potential representations of the n - n interaction , namely the phenomenological potential of tabakin @xcite , and the separable model of ueda et al.@xcite , which is equivalent to the obep
the wave functions of the @xmath33 nn final states for the tabakin potential are of the form @xmath110 while the separable potential model of ueda _ et al .
_ leads to the @xmath33 nn wave function @xmath111 here , @xmath112 is the phase shift of nn - scattering in the @xmath33 wave , and the functions a , @xmath113 , @xmath114 and @xmath115 depend on the choice of parameters @xmath116 , @xmath117 , @xmath118 and @xmath119 for the two models ( see the appendix ) .
we use the non - standard tabakin potential because at short range , the nn wave functions obtained with this potential differ qualitatively from obep wave functions . in fig .
[ figure : obufig2 ] , the wave functions ( [ fst ] ) and ( [ fsu ] ) for both models are shown at an nn lab - energy of @xmath120=100 mev . the relative wave function of eq .
( [ fst ] ) has a node at distances @xmath121 fm ( a stable position of the node in a large interval of nn - energies produces the same nn - scattering phase shifts as a repulsive core ) .
the two models are phase equivalent , but differ in their off - shell behavior . in the following
we will demonstrate that the results for @xmath28 differ considerably for both models , especially if the dibaryon mass @xmath26 comes close to the @xmath25 threshold .
our approximation for the decay amplitude in eq .
( [ tme ] ) leads to a sum over products of two factors .
the first factor is the so - called overlap integral of the intermediate six - quark configuration with the outgoing two - nucleon state .
the second factor is a shell - model transition matrix element that describes the production of the pion on a single quark in the dibaryon and the subsequent transition to an intermediate six - quark configuration .
both factors can be calculated with the standard fractional parentage coefficient ( f.p.c . ) technique , which was developed for quark - model calculations for example in refs .
@xcite . in this subsection
, we calculate the overlap integral of an intermediate six - quark configuration @xmath122 with the ( antisymmetrized and normalized ) @xmath33 partial wave of the final nn - state introduced in eq .
( [ psinn ] ) @xmath123 , [ f_{ct } ] , lst\!=\!001\rangle \ ; . \nonumber\end{aligned}\ ] ] beginning with eq .
( [ ovrl ] ) we denote from now on the nucleon wave function of eq .
( [ n ] ) of the translationally invariant shell - model as @xmath124 here @xmath125 is the total orbital angular momentum contained in the internal jacobi coordinates @xmath126 , and @xmath69 is the h.o .
size parameter for the three - quark system .
the overlap integral ( [ ovrl ] ) is calculated using the standard f.p.c .
technique for the quark shell model @xcite . for this purpose
we use a f.p.c .
decomposition of the six - quark configuration @xmath127 into two three - quark clusters with cst quantum numbers of baryons @xmath128 .
note , that for this procedure , the size parameter in the decomposition @xmath35 differs from the nucleonic size parameter @xmath69 @xmath129[f_{ct}]lst\!=\!001\rangle = \sum_{b_1(n^{\prime } ) } \sum_{b_2(n^{\prime\prime } ) } \sqrt{\frac{n_{f_x^{\prime}}n_{f_x^{\prime\prime}}}{n_{f_x } } } \ ; u_{\{f\}}^{b_1b_2 } \cdot c_{f_x}^{(n)}(n^{\prime},n^{\prime\prime } ) \nonumber \\ & & \qquad\times\left\{\ , \varphi_{\widetilde{n}\widetilde{l } } ( r,\sqrt{2/3}b_6)\ , y_{\widetilde{l}\widetilde{m}}(\hat{\bf r } ) \ , \left\{b_1((n^{\prime } l^{\prime},b_6)123)\ , b_2((n^{\prime\prime } l^{\prime\prime},b_6)456)\ , \right\}_{st\!=\!01 } \right\}_{l\!=\!0 } \label{fpc}\end{aligned}\ ] ] in expansion ( [ fpc ] ) , @xmath130 is a h.o .
wave function in the relative coordinate @xmath131 of the two baryons with angular momentum @xmath132 and @xmath133 excitation quanta ( @xmath134 is the h.o .
size parameter ) . as usual
, @xmath135 is the dimension of the ir @xmath102 $ ] of the permutation symmetry group @xmath136 for six particles @xcite . @xmath137 and @xmath138 are the dimensions of irs @xmath139 $ ] and @xmath140 $ ] of the subgroups @xmath141 and @xmath142 in the reduction @xmath143 .
the coefficients @xmath144 and @xmath145 are f.p.c . in the cst- and x - subspaces respectively . for simplicity , we omit in eq .
( [ fpc ] ) the indices for the dependence of @xmath144 and @xmath145 on the intermediate young schemes @xmath146 , @xmath147 , @xmath148 , @xmath149 , @xmath150 , @xmath151 , @xmath152 , @xmath153 , @xmath154 , and @xmath155 occuring for our chosen reduction chain of eq .
( [ su ] ) . with the help of ( [ fpc ] )
, we can calculate the overlap ( [ ovrl ] ) .
the three - quark three - quark decomposition is of course the most adequate expansion for projecting onto the nn - channel .
the projection for a given intermediate state @xmath156 @xmath157[f_{ct}]lst=001\rangle\ ] ] receives non - vanishing contributions only from nn - components in eq .
( [ fpc ] ) because non - nucleonic clusterings , such as @xmath158 , are orthogonal to @xmath159 in cst space .
furthermore , the overlap of excited nucleonic clusters , e.g. @xmath160 , with the ground state nucleon @xmath161 can be neglected if we assume that the size parameter @xmath35 of the six - quark configuration ( [ fpc ] ) does not differ considerably from the quark core radius @xmath69 of the nucleon .
in fact , because @xmath162 , the non - zero overlap integral between excited and non - excited nucleons is @xmath163 the sum over all possible terms gives a negligible contribution to the final result because the different terms interfere destructively ( see next section ) . due to these restrictions we are lead to the expression @xmath164 where @xmath165=[6]\\ -\sqrt{\frac{1}{5 } } , & & n=2,&[f_x]=[6]\\ -\sqrt{\frac{4}{5 } } , & & n=2,&[f_x]=[42 ] \end{array } \right.\ ] ] the coefficients @xmath145 are calculated by general methods from the tism ( see , e.g. ref .
the values of @xmath166 are given in table [ table1 ] .
the general rule for calculating f.p.c . in the cst - subspace is the factorization of the value @xmath144 @xcite ( symbolically ) @xmath167 in terms of scalar factors @xmath168 and @xmath169 of clebsch - gordan coefficients of the unitary groups @xmath170 and @xmath171 for the reductions @xmath172 and @xmath173 , respectively ( which are links of the common reduction chain ( [ su ] ) ) . the necessary sfs are tabulated in refs . @xcite . with expression ( [ appr ] ) , the overlap integral of eq .
( [ ovrl ] ) reduces to @xmath174 eq .
( [ rovrl ] ) contains a simple radial integral @xmath175 which can be calculated analytically for a plane wave @xmath176 , as well as for the tabakin f.s.i .
wave functions of eq .
( [ fst ] ) and for the ueda f.s.i .
wave functions given in eq .
( [ fsu ] ) .
results for @xmath177 are listed in the appendix .
the large bracket in eq .
( [ rovrl ] ) involving the ratio of the two h.o.size parameters @xmath178 comes from the overlap of the two nucleon clusters @xmath179 . the shell - model matrix element of the pion - production operator @xmath180 defined in eq .
( [ o ] ) ( the second factor in the decay amplitude introduced in eq .
( [ tme ] ) ) is proportional to the one - particle matrix element of the spin - isospin - flip operator @xmath181 .
the remaining five quarks act as spectators for the transition .
chosing the sixth quark , we write the matrix element for the emission of a @xmath182 : @xmath183_x[2 ^ 21 ^ 2]_{ct}lst\!=\!110,j^p\!=\ ! 0
^ - \rangle \nonumber\\ & & \qquad\qquad = 6\ , \frac{f_{\pi q}}{m_{\pi}}\frac{k}{\sqrt{2e_{\pi}(2\pi)^3 } } \langle(n , b_6)\{\bar f\}\bar l\bar s\bar t\!=\ ! 001,t_z\!=\!1\,|\ , ( \bbox{\sigma}_6 \cdot \hat{\bf k})\tau_6^{(+ ) } \exp \left ( { i\frac{5}{6}{\bf k } \cdot \bbox{\rho}_6 } \right ) \,|\ , d^{\prime}\rangle % s^5\!p\,(b_6)[51]_x[2 ^ 21 ^ 2]_{ct}lst\!=\!110,j^p\!=\ ! 0 ^ - \rangle \ ; .\end{aligned}\ ] ] here , @xmath184 , is the jacobi coordinate , and @xmath185 and @xmath186 are spin and isospin of the sixth quark .
the momentum of the pion is @xmath187 , and the factor 6 in front of the one - particle matrix element contains the summation over all six quarks . the natural choice for a f.p.c .
decomposition is clearly the separation of the last quark @xmath188 ( one - particle f.p.c . ) , which allows to exploit the orthogonality constraints for the five spectator quarks . with the one - particle f.p.c.expansion of the shell - model states , the right - hand side of eq .
( [ smtme ] ) reduces to a sum of one - particle spin - isospin - flip amplitudes with algebraic coefficients : @xmath189_x}^{(1)}(s^4p , s)\ , \ , c_{\bar f_x}^{(2)}(s^4p , p)x_6({\bf k};11m,00)\right .
\nonumber \\ & & \qquad\qquad\qquad\qquad\left .
+ c_{[51]_x}^{(1)}(s^5,p ) \left [ \delta_{n,0}\ , c_{[6]_x}^{(0)}(s^5,s)+ \delta_{n,2}\ , c_{[6]_x}^{(2)}(s^5,2s ) \right]x_6({\bf k};11m , n0 ) \right\ } \nonumber \\ & & \qquad\times 3i\langle\bar s^{\prime},s_6:\bar s\!=\!0 \ , | \ , ( \bbox{\sigma}_6 \cdot \hat{\bf k } ) \ , | \ , s^{\prime},s_6:s\!=\!1,m\rangle \langle\bar t^{\prime},t_6:\bar t\!=\!t_z\!=\!1 \ , | \ , \tau_6^{(+ ) } \ , | \ , t^{\prime},t_6:t\!=\!0\rangle \label{series}\end{aligned}\ ] ] where the functions @xmath190 are spatial integrals @xmath191 here @xmath192 and @xmath193 ( @xmath194 and @xmath195 ) are spin and isospin of the five spectator quarks after the separation of the sixth quark . the first term in curly brackets on the right - hand - side of eq .
( [ series ] ) corresponds to the quark transition from an initial s state to a final p state .
the second term corresponds to the quark transition from an initial p state to a final s or 2s state . in eq .
( [ series ] ) , we use almost the same notation for one - particle f.p.c .
@xmath196 and @xmath197 as before for the three - particle f.p.c.s in eq .
( [ fpc ] ) .
note , that the one - particle f.p.c.@xmath196 in the @xmath198-subspace can be calculated with one - particle scalar factors sf ( cf .
( [ sf ] ) ) , which can be found for example in ref . @xcite .
due to the orthogonality restrictions for the five spectator quarks , the summations over @xmath199 collapse to @xmath200 @xmath201 @xmath202 @xmath203 @xmath204 , and the only non - vanishing elementary spin- and isospin - flip amplitudes in eq .
( [ series ] ) are @xmath205 collecting the shell - model matrix element of eqs .
( [ smtme])-([series ] ) for pion production and the overlap integrals of eq .
( [ rovrl ] ) for all intermediate six - quark configurations , and performing the remaining radial integrals in eqs .
( [ rint ] ) and ( [ spatint ] ) , leads to the following result for the full decay amplitude defined in eq .
( [ tme ] ) @xmath206 \ ; . \label{fullam}\end{aligned}\ ] ] the overlap integrals @xmath207 and @xmath208 can be found in the appendix .
note that the inclusion of overlap terms with excited nucleon configurations in the intermediate six - quark states , originating from the different harmonic oscillator parameters @xmath162 , given in eq .
( [ novrl ] ) leads to a non - essential renormalization factor of the decay amplitude ( [ fullam ] ) @xmath209 in front of the term @xmath207 on the right hand side of eq .
( [ fullam ] ) .
this factor ( [ renormi ] ) can be omitted for small @xmath210 .
the @xmath211 behavior of the decay amplitude is due to ( i ) a factor @xmath212 in the transition opertor of eq.([o ] ) , ( ii ) due to the fact that a p - wave quark is involved either in the initial or the final state of the one - particle transition matrix element on the right - hand - side of the eq .
( [ series ] ) .
we recall that in the case of the @xmath30-isobar decay into the @xmath213 channel , the transition matrix element is proportional only to @xmath214 , corresponding to the @xmath215 term in eq .
( [ o ] ) .
the @xmath211 behavior of the @xmath216 decay amplitude ( [ fullam ] ) leads to a very strong dependence of the @xmath1 decay width on the value of @xmath26 , as we will see in the next section .
the total hadronic decay width of the possible @xmath1 dibaryon @xmath28 contains three partial widths @xmath217 which are equal to each other @xmath218 , when we neglect isospin breaking effects . the partial @xmath219 decay width @xmath220 is defined by the standard expression @xcite @xmath221 where @xmath222 is the relative momentum of the two final protons and @xmath223 is the momentum of emitted pion in the c.m . of the @xmath1 dibaryon .
the @xmath90-function conserves the energy in the decay , while the integration over the momentum conserving @xmath224 has already been exploited @xcite in eq .
( [ gampi ] ) .
the integration over three - particle phase space leads to the following result for the partial @xmath225 decay width @xmath226 ^ 2 q^2\,dq \ ; . \label{result}\end{aligned}\ ] ] here , energy conservation relates the pion momentum @xmath227 to the nn relative momentum @xmath228 via @xmath229 \right\}^{1/2}\ ] ] and for @xmath230 , all available decay energy is converted to kinetic energy in the relative nn - system , and none to the pion @xmath231 .
the calculated decay widths are shown in table [ table2 ] , where we have introduced the abbreviations p.w . ,
t and u. here , p.w . refers to a calculation employing a plane - wave final n - n state ( [ pw ] ) , while t and u refer to calculations using the tabakin @xcite ( t ) and ueda et al .
@xcite ( u ) separable nn potentials for the final state interaction . in parentheses
we give the results obtained in the approximation of using only one intermediate six - quark configuration @xmath36 ( n=0 ) . with
the exception of the results for the ueda nn - potential for sets 1 , 2 and 4 ( for which the @xmath1 mass is 400 650 mev above the @xmath25 threshold ) , the inclusion of all pauli principle allowed intermediate 2@xmath4 shell - model configurations tends to increase the decay width by some 20 30 @xmath232 .
the largest effect is obtained for @xmath1 masses rather close to threshold , exemplarily shown for sets 3 and 5 .
it can be seen from table [ table2 ] and fig .
[ figure : obufig3 ] , that the pionic decay width of the @xmath1 is very sensitive to the dibaryon mass @xmath26 , which determines the available phase space of the three - body @xmath25 decay .
the sensitivity grows dramatically near the @xmath25 threshold ( 2016 mev ) .
if we extrapolate the results of table [ table2 ] to the experimental value of @xmath26 = 2065 mev , we obtain a very strong reduction of @xmath28 as compared with the quite realistic variants ( sets 3 and 5 ) in table [ table2 ] : @xmath233 this strong dependence of @xmath28 on the value of @xmath26 is a consequence of the high power of @xmath234 in the integrand of eq .
( [ result ] ) .
the origin of this @xmath235 behavior ( compared with a @xmath236 behavior in case of the @xmath30-isobar decay ) comes , as explained above , from the necessity to excite ( or de - excite ) a p - wave quark for the production of a pion .
note that for small @xmath237 ( when @xmath26 is close to the @xmath17nn threshold ) the function @xmath238 is linear in the factor @xmath239 and can be written as @xmath240 .
therefore , for small @xmath237 the integral in eq .
( [ result ] ) behaves as @xmath241 .
the second high - power factor in eq .
( [ result ] ) is the scale factor @xmath242 which depends sensitively on the ratio @xmath243 .
however , this sensitivity is considerably reduced by the factor @xmath244 in the integrand .
the product @xmath245 is a quite smooth function of @xmath178 . for @xmath69=0.6
fm this product varies from 0.078 @xmath246 to 0.158 @xmath246 , if @xmath35 varies from 0.6 fm to 1.24 fm . for small @xmath237 , f.s.i .
make an important contribution to the @xmath1 decay width because of the large scattering length in the @xmath33 wave @xmath247 fm .
the f.s.i . enhances the decay width for example by about @xmath248 for set 5 in table [ table2 ] . at the experimental mass @xmath26=2065 mev ,
the hadronic decay width is more than doubled by the final state interaction .
it is interesting that in the case of the tabakin model with a nodal nn wave function at short range , the contribution from f.s.i . is smaller than for the ueda model and can even decrease the width compared to the plane wave result ( cf .
this is a direct consequence of an approximate orthogonality of the nodal wave function of the tabakin model to the projection of the intermediate @xmath36 configuration ( i.e. the h.o .
function @xmath249 ) of eq .
( [ appr ] ) onto the nn channel .
this can easily be seen from fig .
[ figure : obufig2 ] , where both wave functions are shown .
the approximate orthogonality of the functions @xmath249 and @xmath250 in the integrand of eq .
( 29 ) reduces considerably the overlap factor @xmath251 , which gives the dominant contribution to the @xmath1 decay width ( see values in parenthesis in table [ table2 ] ) . as it can be seen in fig .
[ figure : obufig3 ] , the disagreement between the tabakin and ueda models grows with increasing dibaryon mass @xmath26 ( the influence of the large scattering length @xmath252 , which is common for both models , becomes negligible compared to the effect of the larger phase space ) . for sets 1 , 2 and 4 in table [ table2 ] ,
the tabakin model leads again to values of @xmath28 , which are even smaller than @xmath28 in the plane wave approximation neglecting f.s.i .
in this work we have studied the pionic decay of a possible @xmath1 dibaryon within the microscopic quark shell - model .
we use a single - quark transition operator which describes the production of the pion on a single quark .
the dibaryon wave function is given as a single six - quark translationally invariant shell model configuration , which has been found to provide an adequate description of the @xmath1 @xcite .
previous results from @xcite have been improved mainly in three points , leading to a complete calculation in the sense , that i ) the calculation is performed consistently on the quark level , i.e. the final two - nucleon state is normalized and antisymmetrized on the quark level , ii ) all important intermediate six - quark states with nonvanishing overlaps with the final two nucleons are included , and iii ) the strong final state interaction for the two nucleons is taken into account on the basis of the separable tabakin @xcite and ueda _ et al .
_ @xcite nn potentials .
not surprisingly , the small available phase space in the three - body decay is the dominant mechanism for the narrow width of the @xmath1 . the large identity factor ( 15 ) , on the other hand , enhances the results of previous evaluations disregarding the identity of quarks from different nucleons .
the inclusion of all pauli principle allowed intermediate 2@xmath4 shell - model configurations tends to increase the decay width by some 20 30 @xmath232 .
furthermore , final state interaction for the two outgoing nucleons also increases the decay width considerably , if the @xmath1 mass is close to the @xmath25 threshold . due to these three effects
, the calculated pionic decay widths lie between @xmath28 = 0.180.32 mev for the most realistic set 5 , having a @xmath1 mass close to the experimentally suggested one and a characteristic hadronic size of the dibaryon of @xmath253 1 fm .
this qualitatively agrees with the experimentally suggested value @xmath28 = 0.5 mev . despite the fact that both , the tabakin and ueda f.s.i .
models , are unsuitable for large nn energies ( as in parameter sets 1 , 2 and 4 ) , the two models demonstrate the strong influence of the short - range behavior of the nn wave function on the @xmath1 decay width .
( see e. g. figs .
[ figure : obufig2 ] and [ figure : obufig3 ] ) . recall that these two models are typical representatives of qualitatively different classes of nn phenomenology .
whereas the ueda separable potential is an approximation of the obep , i. e. a model with short - range repulsion , the tabakin potential can be considered as a unitary - pole approximation ( upa ) @xcite of a moscow - type potential model @xcite with short - range attraction and forbidden states .
the moscow model proceeds from the assumption of a six - quark origin of the short - range nn interaction and pretends to give an adequate description of the non - local character of the nn force .
the main conclusion to be drawn here is , that these two models , which are phase equivalent , differ considerably in their effect on the @xmath1 decay width .
therefore , a possible @xmath1 dibaryon would provide a natural laboratory for detailed studies of the short - range nn interaction .
an interesting continuation of this work would be to go beyond phenomenological nn potential models and use a completely microscopic quark model approach ( see e.g. @xcite and references therein ) . for example , one could calculate the pionic decay of the @xmath1 dibaryon using a final @xmath254 nn - scattering wave function that is based on the same microscopic quark hamiltonian which simultaneously describes the mass @xmath26 and structure of the @xmath1 dibaryon .
however , such a calculation is complicated by the fact that we have used two different hamiltonians , i. e. two different confinement strengths , for three - quark baryons and the six - quark @xmath1 dibaryon @xcite .
thus , the @xmath1 dibaryon could not be explained in terms of the standard constituent quark model , using a common hamiltonian for any number of quarks .
on the other hand , if the @xmath216 really exists , this may be taken as an indication that the effective ( nonperturbative ) quark - quark interaction depends on the state of the system . in this appendix
we present the analytical expressions for the radial integrals @xmath177 defined in eq .
( [ rint ] ) , which are needed to calculate the overlap integral of eq .
( [ ovrl ] ) between different intermediate six - quark shell - model configurations and the two outgoing nucleons .
we recall , that the relative nn - wave function may be described by a simple plane wave ( p.w . ) @xmath176 or by f.s.i .
wave functions resulting , for example , from separable potential representations of the nn - interaction .
we use the results for the nn - wave function obtained by tabakin @xcite given in eq .
( [ fst ] ) and the result obtained by ueda and co - workers @xcite given in eq .
( [ fsu ] ) .
note that coefficients @xmath113 and @xmath255 in eq .
( [ fsu ] ) depend on parameters of the one - term separable potential of ueda et al .
@xcite @xmath256 , @xmath257 . for @xmath113 and @xmath255
we use the following expressions : @xmath258 in table [ table : appendix ] we introduced the following abbreviations : @xmath264 , \\
f^{(0)}(x ) & = & \frac{\sqrt{3}}{x\sqrt{\pi}}-e^{x^2/3 } \left [ 1-\phi(x/3 ) \right ] , \\
f^{(2)}(x ) & = & -\sqrt{\frac{3}{2 } } \left\{\frac{\sqrt{3}}{x\sqrt{\pi } } \left(\frac{1}{3}+\frac{4}{9}x^2\right ) - \left(1+\frac{4}{9}x^2\right ) e^{x^2/3}\left[1-\phi(x/3)\right]\right\ } , \\ g_1^{(0 ) } & = & \frac{\sqrt{3}}{x\sqrt{\pi}}+ \left ( \sin\,x^2/3- \cos\,x^2/3 \right ) \left [ 1-re\phi \left((1+i)x/\sqrt{3 } \right ) \right]- \\ & & \left ( \sin\,x^2/3 + \cos\,x^2/3\right ) im\phi\left((1+i)x/\sqrt{3}\right ) , \\ g_2^{(0 ) } & = & \left ( \sin\,x^2/3 + \cos\,x^2/3\right ) \left[1-re\phi\left((1+i)x/\sqrt{3}\right)\right]+ \\ & & \left ( \sin\,x^2/3- \cos\,x^2/3\right )
im\phi\left((1+i)x/\sqrt{3}\right ) , \\ g_1^{(2 ) } & = & -\sqrt{\frac{3}{2}}\left\ { \frac{\sqrt{3}}{x\sqrt{\pi}}+ \left[\left(\frac{1}{3}+\frac{4}{9}x^2\right ) \sin\,x^2/3- \left(\frac{1}{3}-\frac{4}{9}x^2\right ) \cos\,x^2/3\right ] \left[1-re\phi \left((1+i)x/\sqrt{3}\right ) \right ] \right . \\ & - & \left.\left [ \left ( \frac{1}{3}-\frac{4}{9}x^2\right ) \sin\,x^2/3 + \left ( \frac{1}{3}+\frac{4}{9}x^2\right ) \cos\,x^2/3 \right ] im\phi \left ( ( 1+i)x/\sqrt{3 } \right ) \right\ } , \\
g_2^{(2 ) } & = & -\sqrt{\frac{3}{2 } } \left\ { -\frac{4}{9 } x^2 \frac{\sqrt{3}}{x\sqrt{\pi } } + \left [ \left ( \frac{1}{3}-\frac{4}{9}x^2 \right ) \sin\ , x^2/3 + \left ( \frac{1}{3}-\frac{4}{9}x^2 \right ) \cos\ , x^2/3 \right ] \left [ 1-re\phi \left ( ( 1+i)x/\sqrt{3 } \right ) \right ] \right .
\\ & - & \left .
\left [ \left ( \frac{1}{3}+\frac{4}{9 } x^2 \right ) \sin\,x^2/3 - \left ( \frac{1}{3}-\frac{4}{9 } x^2 \right ) \cos\,x^2/3 \right ] im\phi \left ( ( 1+i)x/\sqrt{3 } \right ) \right\}\end{aligned}\ ] ] * acknowledgments : * + two of the authors ( k.i . and i.t.o . ) acknowledge the warm hospitality extended to them at the institute for theoretical physics , university of tbingen , where this work has been mostly completed .
one of the authors ( k.i . )
wishes to thank the ministry of education , science , sports and culture ( japan ) for financial support .
partially , this work was supported with the rfbr grant no 96 - 02 - 1807 and one of the authors ( i.t.o . )
thanks rfbr for financial support .
thanks the deutsche forschungsgemeinschaft ( dfg ) for a postdoctoral fellowship , contract number wa1147/1 - 1 .
99 r. bilger , b. m. barnet , h. clement , s. krell , g. j. wagner , j. jaki , c. joram , t. kirchner , w. kluge , m. metzler , r. wieser , and d. renker , phys .
b * 269 * , 247 ( 1991 ) ; r. bilger , h. clement , k. fhl , k. heitlinger , c. joram , w. kluge , m. schepkin , g. j. wagner , r. wieser , r. abela , f. foroughi , and d. renker , z. phys .
a * 343 * , 941 ( 1992 ) ; k. fhl , thesis , university of tuebingen 1996 ( submitted for publication ) ; h. clement , m. schepkin , g. j. wagner , and o. zaboronsky , phys .
b * 337 * , 43 ( 1994 ) .
w. a. kaminski , phys .
* 26 * , 148 ( 1995 ) .
b. v. martemyanov and m. g. schepkin , jetp lett .
* 53 * , 776 ( 1991 ) ; r. bilger , h. a. clement , and m. g. schepkin , phys .
71 * , 42 ( 1993 ) ; r. bilger , @xmath213 newsletter * 10 * , 47 ( 1995 ) . h. clement ( private communication ) .
j. mulders , a. t. aerts , and j. j. de swart , phys .
40 * , 1543 ( 1978 ) ; phys .
d * 21 * , 2653 ( 1980 ) .
georg wagner , l. ya .
glozman , a. j. buchmann , and amand faessler , nucl .
* a594 * , 263 ( 1995 ) ; l. ya .
glozman , a. j. buchmann , and amand faessler , j. phys .
g * 20 * , l49 ( 1994 ) .
a. j. buchmann , georg wagner , and amand faessler , submitted to phys . rev .
c ( 1997 ) ; a. j. buchmann , georg wagner , k. tsushima , amand faessler , and l. ya .
glozman , @xmath213 newsletter * 10 * , 68 ( 1995 ) ; prog .
* 36 * , 383 ( 1996 ) .
k. itonaga , a. j. buchmann , georg wagner , and amand faessler , nucl .
a609 * , 422 ( 1996 ) .
k. k. seth , proc .
_ pions in nuclei _ , edited by e. oset ( world scientific , singapore , 1992 ) p. 205 .
f. wang , j .- l . ping , g .- h .
wu , l .- j .
teng , and t. goldman , phys . rev .
c * 51 * , 3411 ( 1995 ) . b. schwesinger and n. n. scoccola , phys .
b * 363 * , 29 ( 1995 ) ; h. garcilazo and l. mathelisch , phys .
* 72 * , 2971 ( 1994 ) ; a. valcarce , h. garcilazo , f. fernandez , and e. moro , phys .
c * 54 * , 1010 ( 1996 ) ; 2085 ( 1996 ) .
f. simkovic and a. faessler , few - body systems suppl .
* 9 * , 231 ( 1996 ) .
m. schepkin , preprint of uppsala univ .
tsl / isv-96 - 01 41 ( 1996 ) .
i. t. obukhovsky , v. g. neudatchin , yu . f. smirnov , and yu .
m. tchuvilsky , phys .
b * 88 * , 231 ( 1979 ) .
m. harvey , nucl .
* a352 * , 301 ( 1981 ) ; 326 ( 1981 ) .
jin - quan chen , j. math .
* 22 * , 1 ( 1981 ) ; j. q. chen , _ group representation theory for physicist _
( world scientific , singapore , 1989 ) ; j. q. chen _ et al .
_ , _ tables of the _ @xmath265 _ coefficients of fractional parentage _ ( world scientific , singapore , 1991 ) .
i. t. obukhovsky , yu . f. smirnov , and yu . f. tchuvilsky , j. phys .
a * 15 * , 7 ( 1982 ) ; i. t. obukhovsky , z. phys . a * 308 * , 253 ( 1982 ) ; prog . part .
phys . * 36 * , 359 ( 1996 ) .
v. g. neudatchin , i. t. obukhovsky , and yu . f. smirnov , phys .
nucl . * 15 * , 519 ( 1984 ) .
s. t. so and d. strotmann , j. math .
* 20 * , 153 ( 1979 ) .
a. m. kusainov , v. g. neudatchin , and i. t. obukhovsky , phys .
c * 44 * , 2343 ( 1991 ) .
a. faessler , f. fernandez , g. lbeck , and k. shimizu , phys .
b * 112 * , 201 ( 1982 ) ; nucl .
* a402 * , 555 ( 1983 ) . f. tabakin , phys .
rev . * 174 * , 1208 ( 1968 ) .
t. ueda , k. tada , and t. kameyama , prog .
* 95 * , 115 ( 1996 ) .
o. f. nemetz , v. g. neudatchin , a. t. rudchik , yu . f. smirnov , and yu . f. tchuvilsky , _ nucleon clusters in atomic nuclei and many - nucleon transfer reactions _ , monograph in russian ( kiev , naukova dumka , 1988 ) ; v. g. neudatchin , yu . f. smirnov , and n. f. golovanova , adv .
11 * , 1 ( 1978 ) .
m. hamermesh , _ group theory and its application to physical problems _ ( adison - wesley , massachusetts - palo alto - london , 1964 ) .
s. nakaishi - maeda , phys .
c * 51*,1663 ( 1995 ) .
v. i. kukulin , v. m. krasnopolsky , v. n. pomerantzev , and b. p. sazonov , phys .
b * 153 * , 7 ( 1985 ) .
a. faessler , a. buchmann , and y. yamauchi , int .
j. mod .
phys . *
2*(1 ) , 39 ( 1993 ) . | we study the pionic decay of a possible dibaryon @xmath0 in the microscopic quark shell model .
the initial @xmath1 dibaryon wave function ( j@xmath2=0@xmath3 , t=0 ) consists of one 1@xmath4 six - quark shell - model @xmath5_x$ ] configuration . the most important final six - quark configurations @xmath6_x$ ] , @xmath7_x$ ] and @xmath8_x$ ] are properly projected onto the nn channel .
the final state nn interaction is investigated by means of two phase - equivalent - but off - shell different - potential models .
we demonstrate that the decay width @xmath9 depends strongly on the short - range behavior of the nn wave function .
in addition , the width @xmath9 is very sensitive to the mass and size of the @xmath1 dibaryon . for dibaryon masses slightly above the experimentally suggested value @xmath10=2.065 gev
, we obtain a pionic decay width of @xmath11 0.180.32 mev close to the experimental value @xmath11 0.5 mev .
= epsf = 16.5truecm = 24.4truecm = -1truecm i.t .
obukhovsky@xmath12 , k. itonaga@xmath13 , georg wagner@xmath14 , a.j .
buchmann@xmath14 + and amand faessler@xmath14 + @xmath15 + @xmath14 + @xmath16 |
being the building blocks of virtually any electronic device , from solar cells to the transistor of integrated circuits , semiconductor materials are the cornerstone of modern technology . despite the wide availability and the low costs of the classical materials like si or gaas ,
alternatives are being sought for applications , involving the use of visible light ( solar energy and optoelectronics ) and/or high voltage drops across the semiconductor ( high - power switches ) .
conductive oxide semiconductors are materials with suitable properties for solar cells or for high - voltage elements , since they have a wide forbidden band - gap and can , at the same time , conduct electricity . while from naive arguments one of these two properties should exclude the other ,
both are present in this material , challenging scientists to explain their behaviour and engineers to apply them in new devices .
one of the open problems is the high conductivity even in nominally undoped materials .
it was attributed to oxygen vacancies , which could offer an alternative conduction channel , as showed in other oxide materials like srtio@xmath1 and its interfaces @xcite .
furthermore , accumulation layers at the surface of the oxides were reported several times in the case of thin films grown on substrates and upon adsorption of water @xcite , which allowed using in@xmath0o@xmath1 as a gas detector .
+ the crystals used in this work are truly bulk single - crystals grown from melt @xcite , and described further below . in this work ,
we analysed the properties of in@xmath0o@xmath1 , a material with a band gap of around 2.7 ev @xcite whose conduction properties have been the subject of debate in the last few years .
numerous theoretical @xcite and experimental @xcite works have been devoted to in@xmath0o@xmath1 , but the microscopic origin of the conduction in this materials remains unclear . to shed light on the problem , we examined the behaviour of the electronic structure of in@xmath0o@xmath1 upon deposition of metallic indium and tin layers measuring the in 4d core levels and valence band .
we observed that the core electron binding energies increased with respect to the clean surface case by the studied metals and a new state appearing near the fermi level in the in@xmath0o@xmath1 forbidden gap at the initial stage of the growth of in .
this state is located at the centre of the brillouin zone , has parabolic dispersion , an effective mass of 0.38 @xmath4 , the latter being the free electron mass , and shows no k@xmath5 dispersion in photon - energy dependent angle - resolved photoemission spectroscopy ( arpes ) measurements .
the intensity of the state increases at low in thicknesses , while the emission intensity fades away at higher coverages .
finally , a uniform metallic - like density of states is present in the whole energy gap of in@xmath0o@xmath1 at higher thicknesses .
metallic sn , possessing one more valence electron , caused a larger downward shift of the bands in comparison to indium , but no new state was measured within the gap .
we exclude that the bottom of the conduction band of in@xmath0o@xmath1 becomes occupied based on the lack of k@xmath5 dispersion .
the new state near the e@xmath6 differs from previously reported electron accumulation layer of the in@xmath0o@xmath1 because of the different effective mass value we found .
for the present study bulk in@xmath0o@xmath1 single crystals were grown from the melt by the levitation - assisted self - seeding crystal growth method @xcite .
the as - grown crystals were dark and almost opaque but turned yellowish - transparent after annealing in oxygen atmosphere for at least 10 hours at 800 - 1000 @xmath7c @xcite .
after this post - growth heat treatment , the crystals showed _
n_-type semiconductor with free electron concentration of 2 @xmath810@xmath9 @xmath10 and electron mobility of 210 @xmath11v@xmath12s@xmath12 according to the hall effect measurements in the van der pauw configuration at room temperature using in - ga ohmic contacts @xcite .
the ( 111)-oriented samples of @xmath13 mm@xmath14 were prepared from large in@xmath0o@xmath1 single crystals to study in and sn / in@xmath0o@xmath1 interfaces using arpes , with ohmic - back contacts made by using silver - epoxy glue .
clean surfaces were obtained by cleavage along the ( 111 ) planes in ultra - high vacuum ( uhv ) .
the normal emission photoelectron spectra were measured by a 5 m normal - incidence monochromated source providing photons in the 5 - 40 ev energy range .
photoelectrons spectra were measured by means of a scienta ses2002 analyzer @xcite .
incident photon energies of 18 , 35 and 39 ev were used to probe the in@xmath0o@xmath1 valence band , the in 4d and sn 4d core levels , respectively .
the measurements were performed at room temperature in a pressure better than 2 @xmath8 10@xmath15 mbar with an energy and angle resolution of 20 mev and 0.2@xmath7 .
the kinetic - energy scale of spectra was calibrated to the fermi level which was known from a thick - polycrystalline indium and tin film on the in@xmath0o@xmath1 crystal .
calibration on a polycrystalline sample yielded the same results .
the metals ( in and sn , purity 99.99 @xmath16 ) were evaporated from a tungsten coil on uhv - cleaved ( 111 ) surfaces of the samples under constant heating conditions at an operating pressure of 5 @xmath8 10@xmath15 mbar keeping the sample at room temperature in a separate vacuum chamber .
the core level and valence band ( vb ) spectra were measured to monitor their band bending by the metal coverage .
figure [ a ] ( a ) shows the core level spectra for different in coverages excited at a photon energy of 35 ev . in fig.[a ] ( b ) , two examples of the fit results for as - cleaved sample as well as after 1 s evaporation of in are shown .
similar to our previous work @xcite and also ref .
@xcite , a fit of the in 4d core level spectra was performed by three gaussians for three components , the spin - orbit split in 4d@xmath17- in 4d@xmath18 doublet and the o - contribution on the lower binding energy side , with the in 4d@xmath17/in 4d@xmath18 intensity ratio and spin - orbit separation was constrained to be 6/4 basing on electron counting arguments , and 0.75 ev .
one can observe a clear shift of the three components of the in-4d core level to the higher binding energy after 1 s of in evaporation .
o@xmath19 versus in coverage at photon energy of 18 ev .
d ) summary of the energy shift of the core level and valence band spectra against in - evaporation time . ]
[ a ] [ ] in addition to the downward band bending of ( 0.44 @xmath20 0.05 ) ev of the core level spectrum , the gradual evolution of two distinct in 4d@xmath17 and in 4d@xmath18 doublet of the in coverage can be noticed due to the metallic non - oxidized indium .
figures [ a ] ( c ) and ( d ) display the arpes spectra of the valence band for the as - cleaved in@xmath0o@xmath1 samples and after various in - deposition steps , and the summary of the downward band bending of the core level and the valence band spectra , respectively . for the valence band spectra ,
the energy of the most intense peak was monitored to determine the band bending . as depicted in figures [ a ] ( c ) and ( d ) , nearly the same downward banding of the valence band spectra with in deposition
was observed .
2 ( a ) shows the selected photocurrent intensity maps along @xmath21 direction of the bulk brillouin zone for the as - cleaved and the low in - covered samples near the fermi energy recorded at @xmath22 ev . after 1 s of in evaporation , electron like state with parabolic dispersion appears .
its intensity increases up to 3 s evaporation time , gradually blurring and disappearing for evaporation times higher than 5 s. figure 2 ( b ) shows the parabola fit of the dispersion of the state in the vicinity of the fermi energy after 3 s in evaporation .
the effective mass , binding energy of the bottom of the band , and the fermi wave vector of a nearly free - electron state were found to be ( 0.38@xmath2 0.05 ) m@xmath3 , ( [email protected] ) ev , and ( [email protected] ) @xmath12 , respectively .
these values differs from those of the quantized subband states which were observed in the near - surface of in@xmath0o@xmath1 thin films @xcite . as indicated in the panel ( c ) of fig .
2 , no photon - energy dispersion of the state near the fermi level can be detected , revealing its two dimensional character .
the surface charge density n@xmath23 is obtained to be 4.08 @xmath810@xmath24 @xmath25 by substitution of the measured fermi wave vector in the formula n@xmath26k@xmath27/2@xmath28 @xcite .
this estimated surface concentration is substantially larger compared to the reported ones at classical semiconductor interfaces @xcite . in spite of the other derived parameters of this state ,
the obtained n@xmath23 is close to that for the clean surface of in@xmath0o@xmath1(111 ) single - crystalline thin film in ref .
the energy difference between bulk fermi level and bottom of the conduction band was determined to be around 0.07 ev in the studied crystals @xcite .
considering the above - mentiond fact as well as the 2d - character of the identified state , we concluded that it is not derived from the conduction band .
the downward band bending of the core level and the valence band spectra in addition to the appearance of the state close to the fermi energy in initial stage of in- deposition , could evidence the electron doping of the surface of the in@xmath0o@xmath1 .
in fact , in might act as a surface donor and lead to downward band bending of the levels and surface fermi level shift to higher binding energies . to test the idea of the surface donor , we evaporated sn , which differs from in for only one electron in the vb .
the selected energy distribution curves ( edc ) in the sn 4d , in 4d core levels , vb and near e@xmath6 region at sn@xmath29in@xmath30o@xmath19 contacts are shown in figures [ c ] ( a ) to ( d ) , respectively .
as shown by spin - orbit split sn 4d doublet upon deposition in panel ( a ) , the metallic sn covers the sample with no sign of formation of sn - o compounds , as expected . from fig .
[ c ] ( b ) and ( c ) , one can see that in comparison to in , sn causes a larger downward band bending of ( -0.58 @xmath20 0.05 ) ev .
this larger shift to higher binding energies can be attributed to stronger donor character of sn with respect to in .
[ c ] ( b ) reveals that in 4d - core level spectra line shape does not change and it is different from that of sn- doped in@xmath30o@xmath19 ( ito ) by sn deposition , ref .
@xcite . therefore , ito formation during the growth of sn on in@xmath30o@xmath19 single crystals can be discarded .
sn coverage results in the suppression of the in 4d states .
the valence band line - shape changes upon tin deposition .
the vb of sn is nearly located in the same energy region .
in contrast to the in@xmath29in@xmath30o@xmath19 , no state within the band gap could be identified in sn - covered samples , as shown in fig .
[ c ] ( d ) .
in@xmath30o@xmath19 interfaces , two distinct spin - orbit sn 4d@xmath17-sn 4d@xmath18 doublet of the metallic sn are observable at binding energies of 24 and 25 ev for high coverage of tin .
b ) , c ) and d ) selected arpes spectra series of in 4d core level , vb and near the fermi energy regions of the clean in@xmath30o@xmath19(111 ) surface and after different evaporation times , respectively . ] [ c ] the work functions of thick layer of in and sn on the in@xmath30o@xmath19 samples were measured _ in situ _ and determined to be ( 4.02 @xmath20 0.05 ) ev and ( 4.31 @xmath20 0.05 ) ev , respectively . by applying these values and also the _ in situ_-deduced electron affinity of the as - cleaved studied crystals ( 4.18 @xmath20 0.06 ) ev
, the schottky - mott rule @xcite predicts a barrier height of ( -0.16 @xmath20 0.11 ) ev for in@xmath29in@xmath30o@xmath19 interfaces and ( 0.13 @xmath20 0.11 ) ev for sn@xmath29in@xmath30o@xmath19 contacts .
these results evidenced the disagreement of the present experimental barrier heights and the predicted ones from this rule , especially for the case of sn@xmath29in@xmath30o@xmath19 contacts .
similar to our previous work @xcite and also ref .
@xcite , the laterally - homogeneous barrier height @xmath31 , was determined for both contacts within the metal - induced gap states ( migs ) based models @xcite . by employing the electronegativity of the studied metals in miedema unit @xcite x@xmath32= 3.90 ev and x@xmath33=4.15 ev
, the barrier heights were derived to be -1.12 ev and -1.02 for in and sn@xmath29in@xmath30o@xmath19 contacts , respectively .
these results also confirm ohmic character of the contacts , but the obtained barrier heights via this model are incompatible with experimental ones .
the difference is not as large as former reported metal - in@xmath30o@xmath19 contacts @xcite .
however , it suggests the improvement of the migs model and especially charge neutrality level ( cnl ) concept for in@xmath30o@xmath19 which might be the origin of this discrepancy .
to conclude , by arpes we showed that despite the similarity between in and sn , the sn and in@xmath29in@xmath30o@xmath19 interfaces behave differently .
these metals act as surface donors generating downward band bending .
the obtained barrier height was larger for the case of sn@xmath29in@xmath30o@xmath19 consistent with its valence - electron configuration . at initial stage of in growth ,
the downward band bending is accompanied by an emerging 2d - electron gas , which vanishes at higher coverages .
the deduced information from the band dispersion of the surface state differs from the frequently - reported seal in in@xmath30o@xmath19 .
in contrast to in growth , no surface state was identified upon sn deposition .
the dissimilarities of the electronic structures of the in and sn - in@xmath30o@xmath19 interfaces despite the comparable structure of these metals , in addition to disagreement of the experimental results and the schottky - mott rule prediction indicate the complexity of contact formation mechanism in these interfaces , whose full explanation should be addressed by ab initio theoretical calculations .
this work was conducted at bessy ii .
we are grateful to k. irmscher for the electrical characterization and the staff of bessy for their technical support .
this work was financially supported by the dfg ( german research foundation ) under project no .
ma2371/8 - 1 .
xx w. siemons , g. koster , h. yamamoto , w. a. harrison , g. lucovsky , t. h. geballe , d. h. a. blank , and m. r. beasley , phys .
lett . * 98 * , 196802 ( 2007 ) .
v. brinzari , g. korotcenkov , m. ivanov , v. nehasil , v. matolin , k. maek , and m. kamei , surf .
sci . * 601 * , 5585 ( 2007 ) .
r. l. kurtz , r. stockbauer , t. e. madey , e. roman , and j. l. de segovia , surf .
sci . * 218 * , 178 ( 1989 ) .
z. galazka , r. uecker , k. irmscher , d. schulz , d. klimm , m. albrecht , m. pietsch , s. ganschow , a. kwasniewski , and r. fornari , j. cryst .
growth * 362 * , 349 ( 2013 ) .
z. galazka , r. uecker , r. fornari ; j. cryst .
growth * 388 * , 61 ( 2014 ) .
v. scherer , c. janowitz , a. krapf , h. dwelk , d. braun , and r. manzke , appl .
. lett . * 100 * , 212108 ( 2012 ) .
p. d. c. king , t. d. veal , f. fuchs , c.y .
wang , d. j. payne , a. bourlange , h. zhang , g. r. bell , v. cimalla , o. ambacher , r.g .
egdell , f. bechstedt , and c. f. mcconville , phys .
b * 79 * , 205211 ( 2009 ) .
k. irmscher , m. naumann , m. pietsch , z. galazka , r. uecker , t. schulz , r. schewski , m. albrecht , and r. fornari , phys .
status solidi a * 211 * , 54 ( 2014 ) .
a. bourlange , d. j. payne , r. g. egdell , j. s. foord , p. p. edwards , m. o. jones , a. schertel , p. j. dobson , and j. l. hutchison , appl .
. lett . * 92 * , 092117 ( 2008 ) .
s. lany , a. zakutayev , t. o. mason , j. f. wager , k. r. poeppelmeier , j. d. perkins , j. j. berry , d. s. ginley , and a. zunger , phys .
lett . * 108 * , 016802 ( 2012 ) .
f. fuchs and f. bechstedt , phys .
b * 77 * , 155107 ( 2008 ) .
karazhanov , p. ravindran , p. vajeeston , a. ulyashin , t. g. finstad , and h. fjellvag , phys .
b * 76 * , 075129 , ( 2007 ) .
j. e. medvedeva , and c. l. hettiarachchi , phys .
b * 81 * , 125116 ( 2010 ) .
p. goston , p. erhart , a. klein , and k. albe , j. phys . : condens .
matter * 21 * , 455801 ( 2009 ) .
p. erhart , a. klein , r. g. egdell , and k. albe , phys .
b * 75 * , 153205 ( 2007 ) .
a. walsh , j. l. f. da silva , s .- h .
wei , c. krber , a. klein , l. f. j. piper , a. demasi , k. e. smith , g.panaccione , p. torelli , d. j. payne , a. bourlange , and r. g. egdell , phys .
rev . lett . * 100 * , 167402 ( 2008 ) . c. janowitz , v. scherer , m. mohamed , a. krapf , h. dwelk , r. manzke , z. galazka , r. uecker , k. irmscher , r. fornari , m. michling , d. schmeisser , j. r. weber , j. b. varley , and c. g. van de walle , , new journal of physics * 13 * , 085014 , ( 2011 ) .
p. d. c. king , t. d. veal , d. j. payne , a. bourlange , r. g. egdell , and c. f. mcconville , phys .
* 101 * , 116808 ( 2008 ) . k. h. l. zhang , r. g. egdell , f. offi , s. iacobucci , l. petaccia , s. gorovikov , and p. d. c. king , phys .
lett . * 110 * , 056803 ( 2013 ) .
o. bierwagen , j. s. speck , t. nagata , t. chikyow , y. yamashita , h. yoshikama , and k. kobayashi , appl .
98 * , 172101 ( 2011 ) .
z. galazka , k. irmscher , m. pietsch , t. schulz , r.uecker , d. klimm , and r. fornari , cryst .
. comm . * 15 * , 2220 ( 2013 ) .
l. f. j. piper , l. colakerol , p. d. c. king , a. schleife , j. ziga - prez , p. a. glans , t. learmonth , a. federov , t. d. veal , f. fuchs , v. muoz - sanjos , f. bechstedt , c. f. mcconville , k. e. smith , phys .
b * 78 * , 165127 ( 2008 ) .
m. nazarzadehmoafi , s. machulik , f. neske , v. scherer , c. janowitz , z. galazka , m. mulazzi , and r. manzke , appl .
. lett . * 105 * , 162104 ( 2014 ) .
t. ando , a. b. fowler , and f. stern , rev .
phys . * 54 * , 437 ( 1982 ) . s. m. sze and k. k. ng , _ physics of semiconductor devices _
( john - wiley & sons hoboken , 2007 ) , chap . iii .
von wenckstern , d. splith , f. schmidt , m. grundmann , o. bierwagen , and j. s. speck , apl materials * 2 * , 046104 ( 2014 ) .
w. moench , j. appl .
phys . * 109 * , 113724 ( 2011 ) .
a. r. miedema , p. f. de chtel , and f. r. de boer , physica b & c * 100 * , 1 ( 1980 ) . | the electronic structure of the transparent semiconductor in@xmath0o@xmath1 has been studied by angle - resolved photoemission spectroscopy upon deposition of metallic indium and also tin on the surface of the semiconductor . by deposition of metallic indium on in@xmath0o@xmath1(111 ) single crystals , we detected the formation of a free - electron like band of effective mass ( 0.38@xmath2 0.05 ) m@xmath3 . at low coverages , metallic in shifts the fermi level of in@xmath0o@xmath1 to higher energies and a new electronic state forms at the metal / semiconductor interface .
this state of two - dimensional character ( 2d - electron gas ) is completely responsible for the electrical conduction in in@xmath0o@xmath1(111 ) at the surface region and has a band dispersion , which does not correspond to the previously found surface accumulation layers in this material . despite the similarity of the electronic properties of in and sn , a larger downward banding
was observed by sn coverage , which was not accompanied by the appearance of the surface state . |
an extrasolar planet s disk - averaged emitting temperature changes throughout its orbit , a phenomenon known as thermal phase variations . in this paper
we use a comprehensive global climate model ( gcm ) to compare the thermal phase variations for two very different climates that are known to have existed in the sun s habitable zone ( hz ) : a temperate climate , and a pan - glacial climate from earth s geological past ( `` snowball earth '' ) . after introducing our climate model ( section 2 ) and discussing its time - averaged characteristics ( section 3 )
, we study the periodic radiative forcings and intrinsic climate response ( section 4 ) .
we then consider , in section 5 , how these climate variations manifest themselves to an external observer ; we discuss our results in section 6 .
such a study can help determine whether time - resolved thermal photometry will be as useful a tool for directly imaged exoplanets as it is for the current crop of short - period giant planets .
as @xcite noted , the most direct way to distinguish a temperate climate from a snowball is to measure the planet s bond albedo . by combining an exoplanet s emitting temperature with its bolometric flux ,
one can obtain estimates of its radius and albedo .
the albedo of a snowball planet would be more than twice that of a terrestrial planet , making such a measurement easy for an infrared direct imaging mission ( e.g. , nasa s terrestrial planet finder interferometer , tpf - i , or esa s darwin ; * ? ? ?
* ; * ? ? ?
* respectively ) . however , an albedo estimate is necessary , but insufficient , for understanding a planet s climate . perversely , the end state at the inner edge of the hz is also highly reflective , if venus is any indication .
a climate constraint orthogonal to albedo is planetary thermal inertia , which depends on atmospheric and surface characteristics .
terrestrial planets may be divided into four categories based on albedo and thermal inertia : 1 ) low albedo and low thermal inertia ( mars ) , 2 ) high albedo and low thermal inertia ( snowball ) , 3 ) high albedo and high thermal inertia ( venus ) , and 4 ) low albedo and high thermal inertia ( earth ) .
furthermore , a large thermal inertia for a planet with a modest atmosphere would constitute circumstantial evidence for liquid water oceans . a planet s thermal inertia can be estimated based on its response to external radiative forcing .
@xcite used an analytic energy balance model ( ebm ) , which assumes diffusive heat transport , to explore how the thermal phase variations of terrestrial planets are affected by a planet s eccentricity , obliquity and thermal inertia .
they found strong degeneracies between these factors : even if the planet s orbit is well characterized via astrometry or imaging , the obliquity , equinox phase and thermal inertia can not be uniquely determined without detecting second - order phase variations .
the present study goes beyond the important work of @xcite in the following ways : 1 ) we use a gcm rather than an analytic ebm , allowing us to self - consistently model the three - dimensional atmosphere and ocean dynamics of these planets , 2 ) we test the effect of planetary thermal inertia using temperate and snowball climates , which are realistic and stable on geological timescales , 3 ) we include the effects of diurnal heating and rotational variability , and 4 ) we simulate ten years worth of observations for each planet , allowing us to evaluate the repeatability of these patterns given stochastic weather and inter - annual variability .
finally , our study is complementary to @xcite : we are unable to vary eccentricity or obliquity because of computational expense , but we consider a wide variety of viewing geometries .
it is useful to first define the system geometry , shown in figure [ geometry ] .
an exoplanet s orbital inclination , @xmath2 , denotes how far it is from being in a face - on orbit : a planet with @xmath3 orbits in the plane of the sky , while @xmath4 means an edge - on orbit .
unless a planet is in a perfectly face - on orbit , it exhibits phases : when the planet is closest to us , inferior conjunction , we primarily see the planet s nightside ; when the planet is farthest from us , superior conjunction , we see mostly dayside . a planet on an edge - on orbit
is `` new '' at inferior conjunction , and `` full '' at superior conjunction .
quadrature occurs when we see half of the illuminated hemisphere of the planet ( as with the moon s quarter phases ) .
a planet on a face - on orbit is permanently in quadrature ; for other inclinations , quadrature occurs twice per orbit . , @xmath5 , @xmath6 , or @xmath7 .
orbital inclination , @xmath2 is the angle between the orbital plane of the planet and the sky plane , or equivalently the angle between the orbital angular momentum and the line of sight .
the sub - observer latitude , @xmath5 , is the latitude directly below the observer .
these two angles are related via the obliquity , @xmath6 , and equinox phase , @xmath7 , via equation [ lat_obs ] .
[ geometry],width=302 ] the angle between the planet s rotation axis , @xmath8 , and its orbital axis , @xmath9 is the obliquity , @xmath6 .
the orientation of the rotational axis with respect to an observer is specified by the equinox phase , @xmath7 .
note that equinox phase is related to but not the same as precession phase , the orientation of the planet s spin axis with respect to periastron .
the sub - observer latitude , @xmath5 , is the latitude of the point directly beneath the observer . if the planet s equator is directly beneath the observer , then @xmath10 ( `` equator - on '' viewing geometry ) , while @xmath11 indicates that the observer is located directly above the planet s north pole ( `` pole - on '' viewing geometry ) .
the sub - observer latitude is related to the inclination , planetary obliquity , and equinox phase : @xmath12 where conservation of angular momentum dictates that all of the parameters in the equation are constants , provided that we neglect precession .
an externally heated planetary atmosphere can roughly be thought of as a heat engine .
for example , surface - to - tropopause and equator - to - pole heating gradients on earth lead to steady - state transport of energy : upward and poleward , respectively .
it is the higher - order periodic climate forcings , however , that cause thermal phase variations ( table [ thermal_phases_table ] and figure [ thermal_phases ] ) .
these variations are damped by heat storage ( i.e. , thermal inertia ) and transport ( atmospheric and/or oceanic ) .
lllll diurnal cycle & rotational@xmath13 & @xmath14 & heat storage and zonal transport@xmath15 & edge - on orbit + eccentricity seasons & orbital & @xmath16 & heat storage & always present + obliquity seasons & orbital & @xmath17 & heat storage and meridional transport & pole - on rotation _ diurnal cycle : _ a planet receives flux from its host star during the day but not at night .
this tends to make a planet s day - side warmer than its night side , leading to phase variations : the day - side is most visible near superior conjunction , and the night side is most visible near inferior conjunction .
insofar as a planet has some thermal inertia , the day night temperature contrast is reduced and the hottest time of day occurs in the afternoon rather than at noon .
if the advective timescale is comparable to , or shorter than , the rotational period , then zonal ( e w ) heat transport becomes an important factor for damping the diurnal temperature swings .
the diurnal cycle is most observable for planets on edge - on orbits , and is the sole source of phase variations for a zero - eccentricity , zero - obliquity planet .
if a planet s surface has inhomogeneous emitting temperature ( e.g. , due to patchy clouds ) , then different regions of the planet will have different emitting temperatures at the same local solar time , leading to _ rotational _ variability .
this manifests itself to an external , inertial observer as high - frequency variations in emitting temperature , where the magnitude of variations depends ultimately on the longitudinal albedo and/or thermal inertia contrasts on the planet . while this effect is included in our simulations and could in principle be used to determine a planet s rotation rate @xcite , it would be averaged over for realistic integration times ( @xmath18 hrs ) , and we do not discuss it further in this paper .
_ eccentricity seasons : _ a planet with non - zero orbital eccentricity tends to be hottest near periastron and coolest near apastron . if the planet has some thermal inertia , the seasonal temperature swings are damped , and the maximum temperature occurs somewhat after periastron .
orbital eccentricity affects the entire planet s power budget , and hence affects thermal phase variations seen from any vantage point ; eccentricity effects may be isolated if a zero - obliquity planet is viewed face - on .
_ obliquity seasons : _ non - zero obliquity leads to the northern hemisphere being hotter during northern summer and cooler during northern winter .
obliquity seasons are damped by thermal inertia and meridional ( n s ) heat transport .
insofar as one hemisphere is more visible to the observer ( @xmath19 ) , obliquity seasons cause thermal phase variations .
if the northern and southern hemispheres have very different albedos and/or thermal inertia , the planet s global power budget will exhibit seasonal variations regardless of viewing geometry , much as with eccentricity seasons .
in general , obliquity seasons are most extreme as seen by a pole - on observer .
a planet on a circular , face - on orbit will have thermal variations entirely dictated by obliquity seasons . a terrestrial planet in the hz of its host star @xcite does not necessarily have a temperate climate or harbor liquid water on its surface .
the high reflectivity of ice and snow compared to land and ocean allows an earth - like planet at the exact same distance from its star and with the same atmosphere to exist in either a temperate state , like modern earth , or an icy snowball state @xcite .
it is believed , for example , that earth has on occasion ( @xmath20710 ma , @xmath20635 ma , and possibly at other times ) fallen into a snowball state , during which global oceans completely @xcite or nearly completely @xcite froze over . as a limiting case , we will be considering here a planet with entirely ice - covered oceans .
the proximate mechanism for the initiation of a snowball state is a runaway ice / snow - albedo feedback : replacing ocean with sea ice leads to cooling and more ice formation @xcite . the ultimate cause of the specific snowball events in earth s history is still unknown @xcite .
the high albedo of ice and snow make a snowball state , once formed , extremely stable @xcite .
compared to earth s 4.5 gyr geological history , snowball events have been rare and brief ( millions of years ) , with a duty cycle of order 1% , probably due to the geologically rapid silicate - weathering feedback @xcite .
the globally frozen continents and low temperatures halted silicate weathering and the associated pulldown of atmospheric co@xmath21 , but greenhouse gases kept being resupplied by volcanism and plate tectonics .
the resulting increase in greenhouse warming in conjunction with other effects such as the dirtying of tropical sea ice @xcite snapped earth out of the snowball state .
it is not known how the escape from a snowball state would play out on exoplanets .
possible differences include the presence of a large water reservoir ( wet planets are more prone to runaway snowball than dune planets ; * ? ? ?
* ) , incident stellar flux ( reduced power budgets make planets more susceptible to global glaciations , including the limit of a starless planet , * ? ? ?
* ; * ? ? ?
* ) , host star spectral type ( bluer stars lead to a stronger temperature albedo feedback ; * ? ? ?
* ) , planetary obliquity ( low - obliquity planets are slightly more prone to snowball events : * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , orbital eccentricity ( larger eccentricity slightly increases the propensity of a planet to enter a snowball state : * ? ? ?
* ; * ? ? ?
* ) , planetary rotation and size ( slower rotation or a smaller planet increases meridional transport , making it easier to enter a snowball state ) , and the effectiveness of an extrasolar carbon - silicate cycle ( continent configuration , water levels , nature of mantle convection , etc . ) .
most importantly for our purposes , a snowball climate has relatively little surface thermal inertia compared to a temperate climate : a few meters equivalent depth of water compared to the current temperate climate with its tens of meters equivalent depth of water . trying to distinguish between these climates based solely on thermal phase variations is therefore a good test for the utility of thermal phase variations in constraining the heat capacity of hz terrestrial exoplanets .
to simulate thermal phase variations for a variety of viewing geometries , we use output from the coupled ocean - atmosphere global climate model echam5/mpi - om .
a detailed description of the gcm is given in @xcite and references therein .
we use top - of - atmosphere ( toa ) incident stellar flux , outgoing longwave radiation , and reflected light , each of which is outputted every two hours at each grid point .
specifically , we here use two simulations presented in @xcite .
these simulations use a modern - earth solar constant of 1367wm@xmath22 , pre - industrial atmospheric carbon dioxide ( 278 ppm ) , a modern - earth orbit ( 23.5@xmath23 obliquity , 1.7% eccentricity ) and two large tropical supercontinents that are meant to represent earth s topography 635 ma before present .
the time - averaged albedo and temperature maps of the two models are shown in figure [ average_maps ] .
@xmath24{exo001_yr10_lat0_lon0_simple_average_albedo_map.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_simple_average_albedo_map.eps}\\ \includegraphics[width=84mm]{exo001_yr10_lat0_lon0_average_t_eff_map.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_average_t_eff_map.eps } \end{array}$ ] the `` temperate '' simulation is initiated from a climate state with polar ice caps and remains in such a state throughout .
the `` snowball '' simulation is initiated from a climate state with globally ice - covered oceans and remains in such a state throughout . as found in previous snowball climate simulations , the low latitudes , and in particular the majority of the continental area ,
remain bare because of a reduced hydrological cycle and net evaporation ( e.g. , * ? ? ?
both simulations were run to equilibrium , then for an additional ten years , with shortwave and longwave radiation outputs saved every two hours .
although our temperate simulation is roughly commensurate with the modern - day earth , it differs in two important ways : 1 ) the continental configuration is much more n s symmetric , reducing the impact of obliquity on the global power budget , and 2 ) the continents are more uniform longitudinally , reducing the amplitude of rotational variations ( cf .
these differences make the resulting climate variations of our temperate planet , if anything , more straightforward to interpret than the modern earth , strengthening our ultimate conclusions .
the time - averaged , total absorbed power of the planet must equal its emitted thermal power .
we verify that this is indeed the case , to one part in a thousand , for the temperate simulation , while the snowball s emitted flux exceeds its absorbed flux by 5.5% .
more energy leaves the atmosphere at the top than enters at the bottom , yet the snowball planet does not cool off , implying that there is a small spurious heat source in the model atmosphere .
this discrepancy , which may be a surface diagnostic bug rather than a true model error , was noted by @xcite and should not significantly affect our results , since we only compare _ time variations _ in thermal emission .
we simply renormalize the snowball simulation s thermal emission to the time - averaged absorbed power in sections 3 & 4 of this paper .
the modest obliquity of our simulated planets dictates that high latitudes receive less orbit - averaged insolation than low latitudes . in figure [ toa_radiative_budget ]
we show the top - of - atmosphere radiative budget for the two simulations . averaging over the seasonal variations that will be the subject of the bulk of this paper , the equator
pole gradient in absorbed flux ( black lines ) results in a meridional temperature gradient , which drives poleward energy transport . as a result , the equator
pole gradient in thermal emission ( red lines ) is muted compared to the gradient in absorbed flux .
@xmath24{exo001_yr10_lat0_lon0_toa_radiative_budget.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_toa_radiative_budget.eps } \end{array}$ ] comparing the radiative budgets of the two simulations , it is clear that the meridional radiative forcing gradient is somewhat greater for the temperate case than for the snowball ( 240 vs. 200 w / m@xmath25 ) , while the resulting gradient in thermal emission is similar for the two planets ( 80 vs. 70 w / m@xmath25 ) .
this indicates that more poleward heat transport is occurring on the temperate planet .
figure [ meridional_transport ] shows the orbit - averaged northward heat transport for the two simulations ; heat transport on the temperate planet is roughly twice that on the snowball .
some of this difference can be attributed to the weaker meridional forcing on the snowball planet .
moreover , transport is much greater in the temperate climate because of two compounding effects : the presence of ocean transport and greater atmospheric humidity . on their own , neither the absence of ocean heat transport , nor the drier air in principle need to grossly affect the total meridional heat transport of an earth - like planet due to `` compensations '' noted and explained by @xcite , @xcite , and @xcite .
but in combination , these two changes make it impossible for the climate to fully compensate , with the net effect that the mid - latitude meridional heat transport in the snowball simulation is significantly reduced .
we now consider periodic changes in the planet s power budget .
we first develop a simple model for interpreting time - variations in absorbed and emitted power , then consider each of the periodic climate forcings in turn , as they manifest themselves in our simulations .
the planet s response to radiative forcing is independent of an external observer , but is the ultimate cause of the thermal phase variations discussed in section 5 . in order to quantitatively
estimate thermal inertia from periodic variations in absorbed and emitted power , we approximate the climate system with the following differential equation ( e.g. , * ? ? ?
* ) : @xmath26 where @xmath27 is the effective thermal inertia coefficient , @xmath28 is the emitting temperature , @xmath29 is the semi - amplitude of variations in the absorbed power , @xmath30 is the driving frequency , and @xmath31 linearly approximates the relation between temperature and emitted flux ( the exact differential equation , analytic and numerical solutions are developed for a blackbody in * ? ? ? * ) .
it is worth noting that this formalism is usually adopted for _ surface _ temperature variations .
nonetheless , the most accessible temperature for any planet other than earth is the _ emitting _ temperature so it is natural to consider effective thermal inertia as it affects top - of - atmosphere absorbed and emitted flux . inferring bulk thermal inertia for the underlying atmosphere and surface
is of course complicated when one is only sensitive to the emitting temperature .
for example , the role of greenhouse gases in equation [ diff_equ ] is normally to reduce @xmath31 , since for a given change in surface temperature , less flux escapes to space . in the current scenario where @xmath28 denotes top - of - atmosphere emitting temperature
, the emitted flux is @xmath32 , where @xmath33 is the stefan - boltzmann constant . in the limit of small temperature excursions
, we may adopt @xmath34 , where @xmath35 is the average emitting temperature .
the greenhouse effect is therefore folded into @xmath27 : greater infrared opacity increases the atmosphere s radiative timescale and hence the lag time between radiative forcing at the surface and response at the top - of - atmosphere .
insofar as upward energy transport in the troposphere is primarily accomplished via convection , the atmosphere s infrared opacity determines the pressure of the emitting layer .
either way , the net result of increasing the concentration of infrared absorbers is increasing the effective thermal inertia coefficient .
the periodic solution to equation [ diff_equ ] is @xmath36 where the temperature oscillations have a semi - amplitude @xmath37 , the phase offset is given by @xmath38 , and the thermal relaxation time is @xmath39 . in this linearized form ,
the emitted flux is proportional to temperature , @xmath40 , so one can estimate the radiative relaxation time based on the amplitudes of variations in absorbed and emitted power : @xmath41 the incoming shortwave radiation undergoes a 6.8% peak - to - trough orbital modulation because of earth s 1.7% eccentricity , while the albedo exhibits seasonal variations of 6% and 4% in the temperate and snowball climates , respectively .
the albedo fluctuations are primarily driven by meridional ( n s ) differences in albedo coupled with seasonal variations in the sub - solar latitude @xcite .
seasonal changes in cloud cover may also contribute to the changing albedo .
this effect can not be predicted _ a priori _ but is an emergent phenomenon of the gcm .
@xmath24{exo001_yr10_lat0_lon0_total_thermal.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_total_thermal.eps } \end{array}$ ] the combination of changing incident flux at the top of the atmosphere and changing albedo leads to peak trough variations in absorbed power of 6% and 8% for the temperate and snowball climates , shown with black lines in figure
[ planet_averaged ] ( in physical units , the two simulations experience approximately the same 7 pw forcing ) .
the changes in absorbed flux drive `` global '' seasons .
the critical difference between global seasons and the usual seasons we experience on earth are that the seasons are in phase in the northern and southern hemispheres , so cross - equatorial heat transport can not damp them .
insofar as the typical eccentricity of exoplanets is 0.10.3 @xcite , it is possible that eccentricity - driven global seasons will be the norm on other planets .
the thermal inertia of the planet damps the amplitude of the oscillations and produces a phase lag , as seen in the planet - integrated emitted power , shown in red in figure [ planet_averaged ] ( peak trough seasonal changes of 1% and 3% for the temperate and snowball models , respectively ) .
note that the low - amplitude response of the temperate planet makes the phase offset difficult to quantify .
for the temperate planet , peak emission occurs nearly simultaneously with the peak absorption , but the minimum emission lags the minimum in forcing by many months . using equation [ tau ] we obtain @xmath42 days and 145 days for the temperate and snowball climates , respectively .
the temperate planet therefore has a thermal relaxation time 2.4@xmath0 longer than that for the snowball .
note that these are longer than the _ atmospheric _ thermal relaxation times because we are accounting for surface heat capacity .
it is often convenient to express the thermal inertia coefficient , @xmath27 , as an equivalent mixed layer depth ( mld ) of water , since oceans are the dominant source of thermal inertia on earth : @xmath43 , where @xmath44 and @xmath45 are the density and specific heat capacity of water . using the globally - averaged emitting temperatures of the two simulations ( @xmath46 k and 232 k )
, we estimate that the temperate and snowball planets have mlds of 25 and 8 meters , respectively ; i.e. , the temperate planet has approximately 3@xmath0 the thermal inertia of the snowball planet .
the greater thermal inertia of the temperate planet is primarily due to the presence of surface liquid water .
following @xcite , earth s oceans have an average mixed layer depth of 2050 m , solid surfaces ( continent or sea ice ) have effective mlds of 23 m , and the atmosphere contributes another 23 m. accounting for the 57% coverage of surface liquid water in our temperate model , one expects it to have 2.45.5 times greater thermal inertia than the snowball planet , in good agreement with our simulations . in figure [ obliquity_seasons ]
we again consider total absorbed and emitted flux , but separate the northern and southern hemispheres in order to isolate the effects of obliquity .
this approach tends to underestimate the magnitude of the seasons , because most of the flux is emitted from low - latitude regions that only experience small seasonal changes in incident radiation .
in fact , the deep tropics receive the most insolation at equinoxes , rather than summer solstice
. nevertheless , the seasonal forcing and response are clearly visible .
the obliquity seasons are approximately @xmath47 greater for the snowball than for the temperate climate ( 26% vs. 4% , as shown by the red lines in figure [ obliquity_seasons ] ) .
if we treat the northern and southern hemispheres independently , ie : ignoring cross - equatorial heat transport , we may again use equation [ tau ] to convert the absorbed / emitted amplitudes into thermal relaxation times .
we obtain values of @xmath48 and 155 days for the temperate and snowball climates , respectively .
since the global and obliquity seasons have the same forcing frequency , one expects the thermal inertia damping for both sorts of seasons to be the same .
the large inferred value of @xmath49 for the temperate simulation obliquity seasons is a testament to the importance of heat transport from the summer to the winter hemispheres .
furthermore , since the obliquity seasons are so much weaker in the temperate simulation , one might conclude that its cross - equatorial heat transport is much greater than that of the snowball planet , but this is not the case .
@xmath24{exo001_yr10_lat0_lon0_total_hemispherical_thermal.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_total_hemispherical_thermal.eps } \end{array}$ ] in figure [ seasonal_transport ] we plot the seasonal cross - equatorial northward heat transport in both simulations . for the temperate planet ,
the ocean and atmosphere both play a role , while the frozen ocean in the snowball climate dictates that its heat transport is entirely atmospheric . despite the lack of ocean transport ,
the seasonality of cross - equatorial energy transport is similar in absolute units for the two simulations .
the temperate planet has a somewhat larger seasonal radiative forcing ( 40 pw vs. 30 pw ; figure [ obliquity_seasons ] ) and much greater annual - mean heat transport than the snowball planet ( cf .
figure [ meridional_transport ] ) . in the absence of thermal inertia
, one would therefore expect the temperate planet to transport much more energy from summer to winter hemisphere than the snowball .
but the greater thermal inertia of the temperate planet leads to a smaller cross - equatorial temperature gradient .
the greater heat transport efficiency and smaller temperature gradient conspire to produce nearly identical cross - equatorial energy transport in the temperate and snowball simulations .
the obliquity seasons have a much larger amplitude for the snowball planet than for the temperate planet , but the phase lag is nearly the same ( 12 weeks ) in both cases .
this is partially due to the qualitatively different effects of heat storage vs. heat transport in damping obliquity seasons : heat storage tends to increase the phase lag while meridional heat transport does not ( this is manifestly true if heat transport is strictly diffusive , as in * ? ? ?
the other reason that obliquity seasons can have very different amplitudes with similar phase lags is the inhomogeneous surface on the temperate planet .
this is important for our temperate simulation because the equatorial landmass and polar ice conspire to leave only 57% of the surface covered in liquid water .
we find that the top - of - atmosphere thermal emission over sea ice exhibits large seasonal temperature variations that are approximately in phase with the forcing , while that over ocean exhibits muted seasonal variations . as noted by @xcite , mixing these signals results in a low amplitude , small - offset response , because the large amplitude component ( solid surfaces ) dominate the offset .
since the land in our simulation is mostly tropical , its toa thermal emission peaks during the relatively dry winter months , while the high surface temperature is masked by overlying clouds and humidity in the summer . to isolate the diurnal cycle , we average over the entire planet and entire orbit , considering absorbed and emitted flux as a function of local solar time , shown in figure [ diurnal_maps ] . for an earth - like planet , zonal ( e w )
heat transport is slow compared to planetary rotation , so the diurnal cycle is primarily damped by the planet s thermal inertia .
@xmath24{exo001_yr10_lat0_lon0_average_olr_zenith_angle_profile.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_average_olr_zenith_angle_profile.eps } \end{array}$ ] we again use equation [ tau ] and obtain thermal relaxation times of @xmath50 and 9 days for the temperate and snowball climates , respectively ( or , equivalently , mlds of 1.1 and 0.5 meters , respectively ) .
these heat capacities are much smaller than those inferred from seasonal forcing because much less material senses the diurnal cycle . for land and sea ice ,
downward heat transport is conductive , so high frequency radiative forcing does not penetrate as deep as longer - period forcing , @xmath51 @xcite .
upward heat transport in the atmosphere and downward heat transport in liquid water are not , strictly speaking , diffusive , but nonetheless only a relatively thin layer of the atmosphere and a shallow layer of the ocean respond to , and damp , the diurnal cycle .
furthermore , the poor vertical ice resolution in our gcm tends to exaggerate the diurnal cycle in surface temperatures @xcite .
this will most affect the snowball simulation , leading us to underestimate its thermal inertia .
the diurnal radiative forcing peaks at noon . in the high - frequency forcing limit
, we expect phase - lags of @xmath52 and @xmath53 for diffusive and mixed - layer heat storage , respectively ( e.g. , section 7.4.2 of * ? ? ?
the actual phase lag is somewhere in between these values .
we now calculate time - resolved , disk - integrated photometry seen by a distant observer , `` light curves '' . to compute the thermal light curves from the planet we assume diffuse emission and neglect limb - darkening : @xmath54 where @xmath55 is the planetary radius , @xmath56 is the top - of - atmosphere outgoing longwave radiation , @xmath57 is latitude , @xmath58 is longitude , @xmath59 is the visibility of a given region of the planet for a given observer ( @xmath60 is unity at the sub - observer point , drops as the cosine of the angle from the sub - observer point , and is zero on the far side of the planet ) , and @xmath61 is the surface integral over a sphere .
viewing geometry only enters equation [ observed_flux ] through @xmath60 , which is a function of the sub - observer location on the planet .
neglecting precession , the sub - observer latitude , @xmath5 , is constant , while the sub - observer longitude evolves linearly in time , @xmath62 , where @xmath63 is the planet s rotational frequency in an inertial frame ( i.e. , the sidereal rather than solar frequency ) . for each viewing geometry we generate 10 years of simulated thermal phase variations and track the root - mean - squared orbit - to - orbit variability .
this inter - annual variability is @xmath64% , ranging between 0.21.5% for most viewing geometries . in figure [ disk_averaged_lat ]
we plot only one year of photometry . for consistency
, we always start the observations at inferior conjunction , when the planet is closest to the observer and we are seeing mostly its night - side .
@xmath24{exo001_yr10_lat0_lon0_observed_fluxes.eps } & \includegraphics[width=84mm]{exo002_yr10_lat0_lon0_observed_fluxes.eps}\\ \includegraphics[width=84mm]{exo001_yr10_lat45_lon0_observed_fluxes.eps } & \includegraphics[width=84mm]{exo002_yr10_lat45_lon0_observed_fluxes.eps}\\ \includegraphics[width=84mm]{exo001_yr10_lat90_lon0_observed_fluxes.eps } & \includegraphics[width=84mm]{exo002_yr10_lat90_lon0_observed_fluxes.eps } \end{array}$ ] the top , middle and bottom panels of figure [ disk_averaged_lat ] show the same simulations , but seen by three different observers .
the short - term variability in the thermal lightcurve is due to regions with different emitting temperatures rotating in and out of view .
these differences in temperature arise because of inhomogeneities in albedo , thermal inertia , and diurnal cloud cycles ( for a thorough discussion of diurnal cycles in earth s olr , see * ? ? ?
since we are considering integration times much longer than the planetary rotation rate , we do not study the rotational variability in this paper .
obliquity seasons are the largest - amplitude response in our simulations , so it is unsurprising that they are the dominant source of thermal phase variations for most viewing geometries .
the obliquity - driven phases are most extreme for a pole - on observer ( top panels of figure [ disk_averaged_lat ] ) and most muted for an equatorial observer ( bottom panels of figure [ disk_averaged_lat ] ) .
this is completely different from the inclination - dependance of diurnally - driven thermal phase variations , which are greatest for edge - on orbits and minimized for face - on orbits ( cf .
* ; * ? ? ?
thermal phase variations are much greater for the snowball than the temperate simulation , as one would expect from section [ obliquity_section ] .
@xcite showed that , in an idealized energy balance model , heat capacity is degenerate with obliquity in thermal phase variations .
we can not test this result because both of our simulated planets have the same obliquity , but the top four panels of figure [ disk_averaged_lat ] shows that the actual situation is likely worse than they concluded .
the peak thermal emission occurs a couple of weeks after summer solstice for both climates , despite the muted amplitude of phase variations in the temperate simulation , because of meridional heat transport and inhomogeneous surface heat capacity ( cf .
section [ obliquity_section ] ) .
in other words , even if one knew the planetary obliquity and solstice phase independently , one would infer an incorrect thermal inertia from equation [ diff_equ ] .
notwithstanding the subtleties of phase lags , the amplitude of thermal phase variations roughly reflect the intrinsic strength of obliquity seasons .
the amplitudes of intrinsic seasons differ by a factor of 6 for the two climates , while the obliquity - driven phase variations differ by 4.5 .
the midlatitude case ( @xmath65 in figure [ disk_averaged_lat ] ) is similar to the polar observer , but with somewhat muted phase variations .
the phase variations seen by a high - latitude observer are dominated by obliquity seasons , but the peak flux roughly coincides with superior conjunction in figure [ disk_averaged_lat ] . if one s intuition is based on thermal phase variations of hot jupiters , it might seem that the maximum flux should always occur at / near superior conjunction , when we are seeing more of the planet s dayside . to dispel this notion
, we plot the lightcurve for a southern observer in figure [ southern_phases ] .
inferior conjunction still occurs at @xmath66 , but now it is the _ southern _ seasons that affect the observed seasonal variability and the planet flux peaks near inferior conjunction . for full caption . ,
width=317 ] for the equatorial view ( bottom panels of figure [ disk_averaged_lat ] ) , the planetary obliquity should not affect the thermal phase variations , since 1 ) the northern and southern hemispheres are equally visible , and 2 ) the continents are nearly equatorial , so there is little to distinguish northern summer from southern summer .
the phase variations from this vantage are therefore driven by global seasons and diurnal heating .
if the albedo of the northern and southern hemispheres were very different , obliquity seasons would affect the phase variations seen by an equatorial observer .
for example , while modern earth s large n
s asymmetry in surface albedo is mostly obscured by the atmosphere and clouds @xcite , the obliquity - driven global seasons are more important than the diurnal cycle for an equatorial observer @xcite .
@xmath67{exo002_yr10_lat90_lon-47_observed_fluxes.eps}\\ \includegraphics[width=84mm]{exo002_yr10_lat90_lon45_observed_fluxes.eps}\\ \includegraphics[width=84mm]{exo002_yr10_lat90_lon135_observed_fluxes.eps } \end{array}$ ] figure [ equatorial_phases ] shows light curves for a snowball climate .
we consider equatorial observers ( @xmath68 ) and fix the inclination at @xmath69 , but vary the argument of periastron , which affects the amplitude and phase offset of thermal phases because of the competing effects of global seasons and diurnal heating ( cf .
we gray - out the phases for which the planet would be inside the inner working angle ( iwa ) of a high - contrast imaging mission and therefore unobservable .
the iwa depends on the details of a mission s design and is a function of wavelength and distance .
we adopt an iwa of 34 mas and an earth analog at 10 pc , so that phase angles within 20@xmath23 of full or new phase or inaccessible ( this is a reasonable iwa for tpf - i ; * ? ? ?
the planet is visible for the entire orbit if its orbital inclination is less than 70@xmath23 , as is the case for the geometries in figures [ disk_averaged_lat ] and [ southern_phases ] . the changing power budget of the planet due to its non - circular orbit leads to a maximum in thermal emission near periastron ( figure [ planet_averaged ] ) .
diurnal heating , on the other hand , leads to a maximum in thermal emission between waxing quarter and superior conjunction , when afternoon local solar times of the planet are most visible ( figure [ diurnal_maps ] ) .
note that earth has westward equatorial winds , but they are slow compared to the planet s rotation , so the hottest local solar time is still in the afternoon when averaged over the visible disk . for planets with thicker atmospheres , the wind direction becomes more relevant than the sense of the planet s rotation ( e.g. , super - rotation on venus , titan , or hot jupiters ) . the thermal phase variations are therefore maximized when superior conjunction occurs a few months after periastron passage , so that the diurnal and eccentricity effects add up ( top panel of figure [ equatorial_phases ] ) ; if the two effects are perfectly out of phase , the phase variations are attenuated ( bottom panel ) . in all cases
the phase variation amplitude is approximately twice as large in the snowball than the temperate simulation , as one would expect from section [ climate_forcing ] .
the diurnal phase variations are smaller than the intrinsic diurnal cycle : even though a single local solar time is directly facing the observer at a given time , the disk - integrated lightcurve averages over many local times .
quantitatively , the first - order diurnal mode is attenuated by @xmath70 with respect to the time - constant component for an edge - on observer @xcite .
since eccentricity affects the entire planet s power budget , it does not suffer from this dilution .
therefore , while the intrinsic diurnal cycle is nearly twice the amplitude of the global seasons , the diurnal contribution to the phase variations is only @xmath71% greater than that from eccentricity effects .
measuring thermal phase variations of an earth analog will require space - based interferometry ( e.g. , nasa s terrestrial planet finder interferometer or esa s darwin mission ; * ? ? ?
* ; * ? ? ?
* respectively ) .
@xcite estimated that _
darwin _ could obtain 20% photometry on an earth - sized planet at 10 pc in 1224 hrs of integration . extrapolating to month - long integrations ,
one expects 2.63.6% photometry for an earth - analog at 10 pc .
we have simulated nine years of photometry , and found the inter - annual variability in disk - integrated flux is @xmath64% , and a mere @xmath72% after averaging over rotational variability .
this opens the possibility of combining multiple orbits - worth of data .
for example , combining the nine years of photometry simulated here provides a three - fold enhancement in the signal - to - noise ratio , or 1% photometry with 1-month integrations . by using bolometric longwave radiation as a proxy for measured flux
, we are assuming an instrument capable of capturing most of the planet s spectral energy distribution ( sed ) .
this is a reasonable assumption because one will need to measure a directly imaged planet s sed in order to estimate its albedo and radius . considering bolometric flux also minimizes the impact of any unseen satellites , which only dominate the system flux in narrow wavebands @xcite .
nevertheless , the presence of a large satellite would considerably complicate the interpretation of thermal phase variations .
when estimating the ability of tpf - interferometer to establish a planet s orbit and residence in the hz , @xcite assumed the planet has constant mid - infrared brightness throughout each orbit .
we find that the thermal flux from an earth analog exhibits orbital variations of 550% .
the amplitude of these variations is primarily governed by the planet s obliquity and the sub - observer latitude , neither of which will be known _ a priori _ for directly imaged planets .
this will complicate the retrieval of their orbital parameters from direct imaging .
note that the situation is yet worse at optical wavelengths : reflected flux variations are of order 100% , since the planet s night side reflects no light .
there is an interesting parallel between non - transiting hot jupiters and directly - imaged planets ( in what follows we assume a circular orbit , but the argument can be generalized to non - zero eccentricity ) . in the hot jupiter scenario ,
the obliquity is assumed to be zero because of tidal effects , but orbital inclination is unknown .
the thermal phase variations are entirely driven by diurnal heating , and therefore put a joint constraint on the planet s size , orbital inclination , and day night temperature contrast ( e.g. , * ? ? ?
* ; * ? ? ?
for directly imaged planets , the observer latitude is unknown , but the orbital inclination can be measured from astrometry . for all but edge - on orbits , thermal phase variations can be assumed to be obliquity - driven , and therefore place a joint constraint on observer latitude and summer winter temperature contrast .
focusing now on directly - imaged planets , the observer latitude is a function of one known quantity , the inclination , and two unknowns : the obliquity and equinox phase ( equation [ lat_obs ] ) . the summer
winter temperature contrast is a function of obliquity , thermal inertia and cross - equatorial heat transport .
two measured quantities , thermal phase amplitude and offset , are clearly insufficient to break the four - way degeneracy . even without breaking any degeneracies , thermal phase variations of an imaged hz planet might still be useful .
small thermal phase variations would improve the odds of habitability , indicating either very weak radiative forcing , high thermal inertia and/or vigorous meridional heat transport .
this test does not work for a planet in a face - on circular orbit , which could simply have a small obliquity .
furthermore , as stated in the introduction , large thermal inertia is no guarantee of habitability ( e.g. , venus ) .
if the planet s obliquity and equinox phase were known through other means ( see section [ constraining_climate ] ) , then one could unambiguously estimate the summer and winter temperatures of the planet .
the seasonal range in temperatures would be a useful climate constraint in and of itself , but one would like to constrain the intrinsic quantities . for a uniform planet , one could break the two - way degeneracy between thermal inertia and heat transport , since they have different phase offset signatures .
but this logic does not hold for an inhomogeneous planet because the mixing of signals from high and low thermal inertia regions complicates the interpretation of the thermal phase amplitude and offset . in principle
it is possible to constrain inhomogeneous surfaces from rotational thermal variability , but this requires higher cadence and therefore a larger mission .
insofar as the rotation rate of a terrestrial planet is determined by the final few major collisions of planet growth , it is natural to expect that some planets will rotate more slowly than earth .
furthermore , planets on circular orbits about low - mass stars will have their planetary rotation rate slowed by tidal forces . in either case
, the result would be stronger diurnal cycles than those considered in this paper .
finally , many directly - imaged planets may have zero obliquity and hence no obliquity seasons .
we have so far focused on true earth analogs , terrestrial planets orbiting in the hz of sun - like stars .
but it is worth noting that , due the compact inner - working angle of an interferometric direct - imaging mission , many of its target stars will be k- and m - dwarfs .
terrestrial planets in the hz of such stars are likely to have had their obliquity tidally eroded @xcite and therefore exhibit thermal phase variations driven only by eccentricity seasons and the diurnal cycle .
such lightcurves would be much easier to interpret and might allow for a relatively robust determination of planetary heat capacity .
we stated in the introduction that a planet s thermal inertia is an attractive `` second parameter , '' after albedo .
we now expand on this idea , describing how various direct imaging observations constrain different key climate parameters , and how thermal phase measurements fit in the big picture . _
0-dimensional model : _ the simplest possible climate model balances absorbed stellar flux with thermal emission , yielding a planet s emitting temperature .
this requires knowing a planet s albedo , which could be extracted from time - averaged thermal observations , but not from reflected - light observations because of the albedo - radius degeneracy .
this is of course insufficient to constrain surface conditions and hence habitability . _
1-dimensional model : _ to estimate surface temperature , one needs to know the pressure of the emitting layer and surface , as well as the atmospheric lapse rate , the rate at which temperature increases as one descends in the troposphere .
a thermal mission will detect any greenhouse gases since by definition they absorb in the thermal ir , and could measure surface pressure via pressure - broadened absorption features @xcite .
an optical mission may be able to detect some greenhouse gases in the near - infrared .
significantly , a thermal mission can directly probe the surface temperature in opacity windows , providing an estimate of lapse rate and a check on climate models .
optical data will contain , at best , indirect information about the planet s temperature .
_ higher - order model : _ to construct a more comprehensive climate model , many additional planetary characteristics must be known .
an optical mission could constrain rotation rate @xcite and obliquity @xcite . furthermore
, the presence of oceans could be inferred via their colors @xcite , polarization @xcite and/or specular reflection ( * ? ? ? * but confounding effects include clouds , robinson et al . 2010 , and snow , cowan et al . 2012 ) .
these quantities would improve our understanding of planet formation and geophysics , and would make it possible to convert observed thermal phase variations into estimates of intrinsic seasonal cycles . for planets theorized to have zero obliquity , thermal phase variations could constrain planetary thermal inertia . but meridional heat transport and surface inhomogeneities would impede efforts to estimate planetary thermal inertia on planets with non - zero obliquity . an exception for which the planetary thermal inertia can be estimated without resorting to periodic climate forcing
is a planet where the thermal inertia is dominated by its atmosphere . recall that the lapse rate is a function of surface gravity and specific heat capacity : @xmath73 , where the _ effective _ specific heat capacity of the planet , @xmath74 , includes the effects of latent heat .
the thermal inertia coefficient for a well - mixed atmosphere is @xmath75 , where @xmath76 is the surface pressure , and the latent heat effects of water vapor are again included implicitly .
a thermal mission of sufficient spectral resolution to measure pressure broadening could therefore estimate an exoplanet s atmospheric thermal inertia , provided that the planetary mass , and hence surface gravity , were otherwise constrained .
each of the three periodic climate forcings diurnal cycle , as well as obliquity and global seasons contribute to the thermal phase variations of habitable zone planets .
the damping for each of these oscillations is different .
the result is that while time - averaged thermal observations will powerfully constrain the climate of directly imaged exoplanets , phase variation measurements will be very challenging to interpret , unless planets can be assured to have zero obliquity . in a few limited cases ,
the lack of thermal variability on orbital timescales would provide circumstantial evidence of clement surface conditions . on the other hand , very large swings in apparent temperature
might discount a planet s habitability .
\1 ) meridional heat transport can significantly affect obliquity seasons , and varies in relative strength from one planet to the next .
surface inhomogeneities in thermal inertia , meanwhile , lead to a mixing of high amplitude , small phase - lag signals with low - amplitude , large phase - lag signals .
these complications make it impossible to quantitatively estimate a planet s heat capacity from obliquity - driven thermal phase variations , even if its obliquity is known through other means .
\2 ) although the rotation rate of an earth analog is shorter than the expected integration times of foreseeable thermal missions , the diurnal cycle shows up in the thermal phase variations on orbital timescales , and is the dominant effect for tropical observers .
for rapidly rotating planets , the diurnal cycle and global seasons are both damped by heat storage alone , but planetary thermal inertia varies as a function of forcing frequency , whether heat storage operates as a mixed - layer or diffusively .
nbc acknowledges an insightful conversation about heat transport on snowball planets with d. koll , and an email exchange about obliquity erosion with r. heller .
av acknowledges support from the german research foundation ( dfg ) program for the initiation and intensification of international collaboration .
this work was supported by the max planck society for the advancement of science .
all simulations were performed at the german climate computing center ( dkrz ) in hamburg , germany . , c. a. 1998 , in society of photo - optical instrumentation engineers ( spie )
conference series , vol .
3350 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
r. d. reasenberg , 719723 | in order to understand the climate on terrestrial planets orbiting nearby sun - like stars , one would like to know their thermal inertia .
we use a global climate model to simulate the thermal phase variations of earth - analogs and test whether these data could distinguish between planets with different heat storage and heat transport characteristics .
in particular , we consider a temperate climate with polar ice caps ( like modern earth ) , and a snowball state where the oceans are globally covered in ice .
we first quantitatively study the periodic radiative forcing from , and climatic response to , rotation , obliquity , and eccentricity .
orbital eccentricity and seasonal changes in albedo cause variations in the global - mean absorbed flux .
the responses of the two climates to these global seasons indicate that the temperate planet has 3@xmath0 the bulk heat capacity of the snowball planet due to the presence of liquid water oceans .
the temperate obliquity seasons are weaker than one would expect based on thermal inertia alone ; this is due to cross - equatorial oceanic and atmospheric energy transport . thermal inertia and cross - equatorial heat transport
have qualitatively different effects on obliquity seasons , insofar as heat transport tends to reduce seasonal amplitude without inducing a phase lag . for an earth - like planet , however , this effect is masked by the mixing of signals from low thermal inertia regions ( sea ice and land ) with that from high thermal inertia regions ( oceans ) , which also produces a damped response with small phase lag .
we then simulate thermal lightcurves as they would appear to a high - contrast imaging mission ( tpf - i / darwin ) . in order of importance to the present simulations ,
which use modern - earth orbital parameters , the three drivers of thermal phase variations are 1 ) obliquity seasons , 2 ) diurnal cycle , and 3 ) global seasons .
obliquity seasons are the dominant source of phase variations for most viewing angles
. a pole - on observer would measure peak - to - trough amplitudes of 13% and 47% for the temperate and snowball climates , respectively .
diurnal heating is important for equatorial observers ( @xmath1% phase variations ) , because the obliquity effects cancel to first order from that vantage .
finally , we compare the prospects of optical vs. thermal direct imaging missions for constraining the climate on exoplanets and conclude that while zero- and one - dimensional models are best served by thermal measurements , second - order models accounting for seasons and planetary thermal inertia would require _ both _ optical and thermal observations . |
we study the gross - pitaevskii equation ( nls ) with a repulsive delta function potential ( @xmath1 ) @xmath2 as initial data we take a fast soliton approaching the impurity from the left : @xmath3 because of the homogeneity of the problem this covers the case of the general soliton profile @xmath4 .
the quantum transmission rate at velocity @xmath5 is given by the square of the absolute value of the transmission coefficient , see below , @xmath6 for the soliton scattering the natural definition of the transmission rate is given by @xmath7 provided that the limit exists .
we expect that it _ does _ and that for fixed @xmath8 , there is a @xmath9 such that @xmath10 based on the comparison with the linear case ( see below ) and the numerical evidence @xcite we expect with @xmath11 . towards this heuristic claim we have [ th:1 ]
let @xmath12 satisfy @xmath13 .
if @xmath14 is the solution of with initial condition and @xmath15 , then for fixed @xmath8 , @xmath16 uniformly for @xmath17 we see that by taking @xmath18 very close to @xmath19 , we obtain an asymptotic rate just shy of @xmath20 .
more precisely , we show that there exists @xmath21 diverging to @xmath22 as @xmath23 and @xmath24 , such that for fixed @xmath25 , if @xmath26 , then @xmath27 the constant @xmath28 appearing here is independent of all parameters ( @xmath29 , @xmath30 , and @xmath18 )
. we have conducted a numerical verification of theorem [ th:1 ] see fig .
[ f : nice ] .
it shows that the approximation given by is very good even for velocities as low as @xmath31 , at least for @xmath32 a more elaborate numerical analysis will appear in our forthcoming paper @xcite .
our second result shows that the scattered solution is given , on the same time scale , by a sum of a reflected and a transmitted soliton , and of a time decaying ( radiating ) term see the fourth frame of fig .
[ f : solv3 ] .
this is further supported by a forthcoming numerical study @xcite . in previous works in the physics literature
( see for instance @xcite ) the resulting waves were only described as `` soliton - like '' .
[ th:2 ] under the hypothesis of theorem [ th:1 ] and for @xmath33 we have , as @xmath34 , @xmath35 where @xmath36 , @xmath37 , and @xmath38 @xmath39 here @xmath40 and @xmath41 are the transmission and reflection coefficients of the delta - potential ( see ) . when @xmath42 or @xmath43 the first error term in is modified to @xmath44 .
here and later we use the standard notation @xmath45 this asymptotic description holds for @xmath30 greater than some threshold depending on @xmath8 and @xmath18 , as in theorem [ th:1 ] . the implicit constant in the @xmath46 error term is entirely independent of all parameters ( @xmath29 , @xmath30 , and @xmath18 ) , although the implicit constant in the @xmath47 error term depends upon @xmath8 , or more precisely , the proximity of @xmath48 and @xmath49 to @xmath50 .
a comparison of the transmission and reflection coefficients of the @xmath12 potential , and of the soliton transmission and reflections coefficients , is shown in figure [ f : lnl ] .
scattering of solitons by delta impurities is a natural model explored extensively in the physics literature see for instance @xcite,@xcite , and references given there . the heuristic insight that at high velocities `` linear scattering '' by the external potential should dominate the partition of mass is certainly present there . in the mathematical literature
the dynamics of solitons in the presence of external potentials has been studied in high velocity or semiclassical limits following the work of floer and weinstein @xcite , and bronski and jerrard @xcite see @xcite for recent results and a review of the subject .
roughly speaking , the soliton evolves according to the classical motion of a particle in the external potential .
that is similar to the phenomena in other settings , such as the motion of the landau - ginzburg vortices .
the possible novelty in and lies in seeing _
quantum _ effects of the external potential strongly affecting soliton dynamics . as shown in fig .
[ f : nice ] , theorem [ th:1 ] gives a very good approximation to the transmission rate already at low velocities .
[ f : solv3 ] shows time snapshots of the evolution of the soliton , and the last frame suggests the soliton resolution . we should stress that the asymptotic solitons are resolved at a much larger time
see @xcite . the proof of the two theorems , given below in [ proof][resolution ] , proceeds by approximating the solution during the `` interaction phase '' ( the interval of time during which the solution significantly interacts with the delta potential at the origin ) by the corresponding linear flow .
this approximation is achieved , uniformly in @xmath29 , by means of strichartz estimates established in [ ros ] .
the use of the strichartz estimates as an approximation device , as opposed to say energy estimates , is critical since the estimates obtained depend only upon the @xmath0 norm of the solution , which is conserved and _ independent _ of @xmath30 .
thus , @xmath30 functions as an asymptotic parameter ; larger @xmath30 means a shorter interaction phase and a better approximation of the solution by the linear flow .
theorem 2 combines this analysis with the inverse scattering method .
the delta potential splits the incoming soliton into two waves which become single _
solitons_. acknowledgments .
we would like to thank mike christ , percy deift , and michael weinstein for helpful discussions during the preparation of this paper .
the work of the first author was supported in part by an nsf postdoctoral fellowship , and that of the second and third author by nsf grants dms-0354539 and dms-0200732 .
here we present some basic facts about scattering by a delta - function potential on the real line .
let @xmath51 and put @xmath52 we define special solutions , @xmath53 , to @xmath54 , using notation given in : @xmath55 where @xmath56 and @xmath57 are the the transmission and reflection coefficients : @xmath58 they satisfy two equations , one standard ( unitarity ) and one due to the special structure of the potential : @xmath59 we use the representation of the propagator in terms of the generalized eigenfunctions see for instance the notes @xcite covering scattering by compactly supported potentials .
the resolvent @xmath60 has kernel given by @xmath61 this gives an explicit formula for the spectral projection , and hence the schwartz kernel of the propagator : @xmath62 the propagator for @xmath63 is described in the following [ l : lin ] suppose that @xmath64 and that @xmath65 $ ]
. then @xmath66 where @xmath67 all we need to do is to combine and . using the support property of @xmath68 we compute , @xmath69 so that @xmath70 where we used the fact that @xmath71 . similarly , using @xmath72 , we have @xmath73 a simple computation gives concluding the proof .
we have two simple applications of lemma [ l : lin ] : the strichartz estimates ( proposition [ p : str ] ) and the asymptotics of the linear flow @xmath74 as @xmath75 ( proposition [ p : as ] ) .
we start with the strichartz estimate , which will be used several times in the various approximation arguments of [ proof ] .
since it is particularly simple in our setting , we give a complete proof ( see @xcite for references and the general version ) .
[ p : str ] suppose @xmath76 and @xmath77 let the indices @xmath78 , @xmath79 , @xmath80 satisfy @xmath81 and fix a time @xmath82 .
then @xmath83 } l^r_x } \leq c \| \varphi \|_{l^2 } + c \| f \| _ { l_{[0,t]}^{\tilde p } l_x^{\tilde r } } \ ] ] the constant @xmath28 is independent of @xmath29 and @xmath84 . moreover , in , we can take @xmath85 and , on the right - hand side of , replace @xmath86}^{\tilde p } l_x^{\tilde r } } $ ] with @xmath87}^\frac{4}{3}}$ ] .
we put @xmath88 , so that @xmath89 is a unitary group on @xmath90 .
for @xmath91 we have , using lemma [ l : lin ] , @xmath92 by the riesz - thorin interpolation theorem ( see for instance ( * ? ?
* theorem 7.1.12 ) ) we have @xmath93 the estimate with @xmath94 reads @xmath95 which by duality is equivalent to @xmath96 the two equivalent estimates together give ( is applied with @xmath97 , @xmath98 replaced by @xmath79 , @xmath80 it is easily checked that still holds ) @xmath99 putting @xmath100 } ( s ) f ( s , x ) $ ] we obtain for @xmath101 .
hence it suffices to prove . put @xmath102 and note that @xmath103 .
the estimate is equivalent to @xmath104 which is the same as @xmath105 to obtain from we apply the hardy - littlewood - sobolev inequality which says that if @xmath106 and @xmath107 then @xmath108 see for instance ( * ? ? ?
* theorem 4.5.3 ) . we apply it with @xmath109 which is the admissibility condition .
we now turn to the large velocity asymptotics of the linear flow @xmath74 .
[ p : as ] let @xmath110 be bounded , together will all of its derivatives .
let @xmath111 , @xmath112 , and suppose @xmath113 \subset ( -\infty,0]$ ] .
then for @xmath114 , @xmath115 = \begin{aligned}[t ] & t(v)e^{-ith_0}[e^{ixv}\varphi(x - x_0 ) ] \\ & + r(v ) e^{-ith_0}[e^{-ixv}\varphi(-x - x_0 ) ] \\ & + e(x , t ) \end{aligned}\ ] ] where , for any @xmath116 , @xmath117 & \frac{1}{v}\|\partial_x [ \theta(x)\varphi(x - x_0)]\|_{l^2}\\ & + \frac{c_k}{(tv)^k } \| { \langle}x { \rangle}^k \varphi(x ) \|_{h^k}\\ & + 4\|(1-\theta(x))\varphi(x - x_0)\|_{l_x^2 } \end{aligned}\ ] ] in [ proof ] , proposition [ p : as ] will be applied with @xmath118 a smooth cutoff to @xmath119 , and @xmath120 with @xmath121 . before proving proposition [ p : as ] ,
we need the following [ p : as0 ] let @xmath122 with @xmath123 $ ]
. then @xmath124(x ) = \begin{aligned}[t ] & e^{-ith_0}[e^{ixv}\psi(x)](x)\ , x_-^0 \\ & + t(v)e^{-ith_0}[e^{ixv}\psi(x)](x ) \ , x_+^0 \\ & + r(v)e^{-ith_0}[e^{-ixv}\psi(-x)](x ) \ , x_-^0\\ & + e(x , t ) \end{aligned}\ ] ]
where @xmath125 uniformly in @xmath126 . by with @xmath127 , @xmath128 & [ e^{-ith_0}(\varphi\ast ( \tau - t(v)\delta_0))(x)]\ , x_+^0 \\
& + [ e^{-ith_0}(\varphi\ast ( \rho - r(v)\delta_0))(-x)]\ , x_-^0 \end{aligned}\ ] ] and thus it suffices to show @xmath129 and @xmath130 the proofs of these two estimates are similar , so we only carry out the proof of . by unitarity of @xmath131 and
the plancherel s identity , @xmath132(x)\|_{l_x^2 } = \| \hat{\psi}(\lambda -v ) ( t(\lambda)-t(v))\|_{l_\lambda^2 } \,.\ ] ] since @xmath133 we have @xmath134 . using this to estimate the right - hand side of and applying plancherel s identity again yields .
apply to @xmath135 to obtain @xmath136(x ) = \begin{aligned}[t ] & e^{-ith_0}[e^{ixv}\varphi(x - x_0)](x)\ , x_-^0 \\ & + t(v)e^{-ith_0}[e^{ixv}\varphi(x - x_0)](x ) \
, x_+^0 \\ & + r(v)e^{-ith_0}[e^{-ixv}\varphi(-x - x_0)](x ) \ , x_-^0\\ & + e_1(x , t)+e_2(x , t ) \end{aligned}\ ] ] where @xmath137 is as in lemma [ p : as0 ] and ( putting @xmath138 ) @xmath139 & + e^{-ith_q}f(x ) - e^{-ith_0}f(x ) \ , x_-^0 \\ & -t(v)e^{-ith_0}f(x)\ , x_+^0 - r(v ) e^{-ith_0}[f(-x)](x ) \ , x_-^0 \end{aligned}\ ] ] by lemma [ p : as0 ] , @xmath140\|_{l_x^2}\ ] ] uniformly for all @xmath126 , and by unitarity of the linear flows , @xmath141 also uniformly in all @xmath126 .
now restrict to the time interval @xmath114 . by , it remains to show that @xmath142(x)\|_{l_{x<0}^2 } \leq \frac{c_k}{(tv)^k}\|{\langle}x { \rangle}^k\varphi(x)\|_{h_x^k } \\
\notag\|e^{-ith_0}[e^{-ixv}\varphi(-x - x_0)](x)\|_{l_{x>0}^2 } \leq \frac{c_k}{(tv)^k}\|{\langle}x { \rangle}^k\varphi\|_{h^k}\end{gathered}\ ] ] the second of these is in fact equivalent to the first , since for any function @xmath143 , @xmath144(x ) = e^{-ith_0}[g(x)](-x)\,.\ ] ] now we establish . since @xmath145 } \ ,
\hat { } \ , ( \lambda ) = e^{-ix_0(\lambda - v)}\hat{\varphi}(\lambda - v ) \,,\ ] ] @xmath146(x ) & = \frac{1}{2\pi } \int e^{ix\lambda } e^{-ix_0(\lambda - v)}e^{-it\lambda^2/2}\hat{\varphi}(\lambda - v)\ , d\lambda\\ & = e^{-itv^2/2}e^{ixv}\frac{1}{2\pi } \int e^{i\lambda(x - x_0-tv)}e^{-it\lambda^2/2}\hat{\varphi}(\lambda ) \ , d\lambda\end{aligned}\ ] ] by @xmath147 applications of integration by parts in @xmath148 , @xmath149 \ , d\lambda\ ] ] since @xmath150 , we have @xmath151 and thus @xmath152 for @xmath119 .
hence @xmath153(x)|^2 \ , dx \nonumber \\ & \leq \frac{c_k}{(tv)^k } \left\| \int e^{i\lambda(x - x_0-tv)}\partial_\lambda^k[e^{-it\lambda^2/2}\hat{\varphi}(\lambda ) ] \ , d\lambda \right\|_{l_x^2 } \\ & = \frac{c_k}{(tv)^k } \left\| \partial_\lambda^k[e^{-it\lambda^2/2}\hat{\varphi}(\lambda ) ] \right\|_{l_\lambda^2 } \nonumber \end{aligned}\ ] ] from which the result follows by applying the leibniz product rule and the plancherel identity once again ( and using that @xmath154 ) . * remark .
* suppose that @xmath155 \ , , \ \
\psi \in { \mathcal s } ( { { \mathbb r } } ) \ , , \
\ { \operatorname{supp}}\psi \subset ( - \infty , 0 ) \ , , \ \
\| \psi \|_{l^2 } = 1 \,.\ ] ] then for @xmath156 and as @xmath157 , @xmath158 in fact using and an estimate similar to we see that for @xmath159 @xmath160 an expansion in powers of @xmath161 gives .
in this section , we prove theorem [ th:1 ] .
we recall the notation for operators from sect.[ros ] and introduce short hand notation for the nonlinear flows : * @xmath162 .
the flow @xmath163 is termed the `` free linear flow '' * @xmath164 .
the flow @xmath165 is termed the `` perturbed linear flow '' * @xmath166 , termed the `` perturbed nonlinear flow '' is the evolution of initial data @xmath167 according to the equation @xmath168 * @xmath169 , termed the `` free nonlinear flow '' is the evolution of initial data @xmath167 according to the equation @xmath170 from sect.[in ] we recall the form of the initial condition : @xmath171 , @xmath172 , @xmath173 , @xmath13 , and we put @xmath174 .
we begin by outlining the scheme , and will then supply the details .
the @xmath175 notation always means @xmath176 difference , uniformly on the time interval specified , and up to a multiplicative factor that is independent of @xmath29 , @xmath30 , and @xmath18 ( any such dependence will be exhibited explicitly ) . * phase 1 ( pre - interaction)*. consider @xmath177 , where @xmath178 so that @xmath179 . the soliton has not yet encountered the delta obstacle and propagates according to the free nonlinear flow @xmath180 the analysis here is valid provided @xmath30 is greater than some absolute threshold ( independent of @xmath29 , @xmath30 , or @xmath18 ) .
but if we further require that @xmath30 be sufficiently large so that @xmath181 ( recall @xmath182 ) , then @xmath183 .
this is the error that arises in the main argument of phase 2 below .
* phase 2 ( interaction)*. let @xmath184 and consider @xmath185 .
the incident soliton , beginning at position @xmath186 , encounters the delta obstacle and splits into a transmitted component and a reflected component , which by time @xmath187 , are concentrated at positions @xmath188 and @xmath186 , respectively .
more precisely , at the conclusion of this phase ( at @xmath187 ) , @xmath189 & t(v)e^{-it_2v^2/2}e^{it_2/2}e^{ixv}{\textnormal{sech}}(x - x_0-vt_2)\\ & + r(v)e^{-it_2v^2/2}e^{it_2/2}e^{-ixv}{\textnormal{sech}}(x+x_0+vt_2 ) \\ & + \mathcal{o}(v^{-\frac{1}{2}\delta } ) \end{aligned}\ ] ] this is the most interesting phase of the argument , which proceeds by using the following three observations * the perturbed nonlinear flow is approximated by the perturbed linear flow for @xmath185 . *
the perturbed linear flow is split as the sum of a transmitted component and a reflected component , each expressed in terms of the free linear flow of soliton - like waveforms . *
the free linear flow is approximated by the free nonlinear flow on @xmath185 .
thus , the soliton - like form of the transmitted and reflected components obtained above is preserved .
the brevity of the time interval @xmath190 $ ] is critical to the argument , and validates the approximation of linear flows by nonlinear flows .
* phase 3 ( post - interaction)*. let @xmath191 , and consider @xmath192 $ ] .
the transmitted and reflected waves essentially do not encounter the delta potential and propagate according to the free nonlinear flow , @xmath193 & e^{-itv^2/2}e^{it_2/2}e^{ixv}{\textnormal{nls}_0}(t - t_2)[t(v){\textnormal{sech}}(x)](x - x_0-tv ) \\ & + e^{-itv^2/2}e^{it_2/2}e^{-ixv}{\textnormal{nls}_0}(t - t_2)[r(v){\textnormal{sech}}(x)](x+x_0+tv ) \\ & + \mathcal{o}(v^{1-\frac{3}{2}\delta } ) , \qquad t_2\leq t \leq t_3 \end{aligned}\ ] ] this is proved by a perturbative argument that enables us to evolve forward a time @xmath194 at the expense of enlarging the error by a multiplicative factor of @xmath195 .
the error thus goes from @xmath196 at @xmath187 to @xmath197 at @xmath198 .
now we turn to the details .
let @xmath199 and @xmath200 .
let @xmath201 . recall that @xmath202 so that @xmath179 .
note that @xmath203 we will need the following perturbation lemma .
[ l : approx1 ] if @xmath204 , @xmath205 , and @xmath206}^\infty } \leq 1 $ ] , then @xmath207}^\infty l_x^2 } \leq c_2(\|w(\cdot , t_a)\|_{l_x^2}+q\|u_1(0,t)\|_{l_{[t_a , t_b]}^\infty})\ ] ] where the constants @xmath208 and @xmath209 depend only on constants appearing in the strichartz estimates and are , in particular , independent of @xmath29 and @xmath30 .
@xmath210 solves @xmath211 from this equation , @xmath210 is estimated using proposition [ p : str ] .
for the cubic nonlinear term we take @xmath212 and estimate by hlder as @xmath213}^{6/5}l_x^{6/5 } } \leq ( t_b - t_a)^{1/2}\|w\|_{l_{[t_a , t_b]}^6l_x^6}^2 \|w\|_{l_{[t_a , t_b]}^\infty l_x^2}\ ] ] since complex conjugates becomes irrelevant in the estimates , both quadratic terms are treated identically . in proposition
[ p : str ] , we take @xmath212 and estimate by hlder as @xmath214}^{6/5}l_x^{6/5 } } & \leq ( t_b - t_a)^{1/2}\|w\|_{l_{[t_a , t_b]}^6l_x^6}^2 \|u_1\|_{l_{[t_a , t_b]}^\infty l_x^2 }
\\ & \leq \sqrt 2(t_b - t_a)^{1/2 } \|w\|_{l_{[t_a , t_b]}^6l_x^6}^2\end{aligned}\ ] ] for the linear terms ( both of the form @xmath215 ) , we take @xmath212 in proposition [ p : str ] and estimate as @xmath216}^{6/5}l_x^{6/5 } } & \leq ( t_b - t_a)^{1/2}\|w\|_{l_{[t_a , t_b]}^6l_x^6 } \|u_1\|_{l_{[t_a , t_b]}^6 l_x^6 } \|u_1\|_{l_{[t_a , t_b]}^\infty l_x^2 } \\ & \leq 2(t_b - t_a)^{2/3 } \|w\|_{l_{[t_a , t_b]}^6l_x^6}\end{aligned}\ ] ] the delta term is estimated by the concluding sentence of proposition [ p : str ] as @xmath217}^{4/3 } } \leq q ( t_b - t_a)^{3/4 } \|u(0,t)\|_{l_{[t_a , t_b]}^\infty}\ ] ] since @xmath218 , collecting the above estimates we have ( taking @xmath219}^\infty l_x^2 } + \|w\|_{l_{[t_a , t_b]}^6l_x^6}$ ] ) , @xmath220}^\infty}\ ] ] provided @xmath221 above , the linear term on the right can be absorbed by the left as @xmath222}^\infty}\ ] ] continuity of @xmath223 as a function of @xmath224 shows that provided @xmath225}^\infty } ) \leq 1/2 $ ] , the above estimate implies @xmath226}^\infty}\ ] ] concluding the proof .
now we proceed to apply lemma [ l : approx1 ] .
the constants @xmath208 and @xmath209 will , for convenience of exposition , be taken to be @xmath227 and @xmath228 .
let @xmath116 be the integer such that @xmath229 .
( note that @xmath230 if the soliton starts within a distance @xmath30 of the origin , i.e. @xmath231 , and the inductive analysis below is skipped . ) apply lemma [ l : approx1 ] with @xmath232 , @xmath233 to obtain ( since @xmath234 ) @xmath235}^\infty l_x^2 } \leq 2q\|u_1(0,t)\|_{l_{[0,1]}^\infty}\leq 2q{\textnormal{sech}}(x_0+v)\ ] ] apply lemma [ l : approx1 ] again with @xmath236 , @xmath237 to obtain @xmath238}^\infty l_x^2 } & \leq 2(\|w(\cdot,1)\|_{l_x^2}+ q\|u_1(0,t)\|_{l_{[1,2]}^\infty } ) \\
& \leq 2 ^ 2q{\textnormal{sech}}(x_0+v)+2 ^ 1q{\textnormal{sech}}(x_0 + 2v)\end{aligned}\ ] ] we continue inductively up to step @xmath147 , and then collect all @xmath147 estimates to obtain the following bound on the time interval @xmath239 $ ] @xmath240}^\infty l_x^2 } \leq 2q \sum_{j=1}^k 2^{k - j}{\textnormal{sech}}(x_0+jv)\ ] ] the estimate @xmath241 reduces matters to bounding @xmath242 and , after summing the geometric series , we obtain @xmath240}^\infty l_x^2 } \leq c2^k e^{x_0+v } \frac{(2^{-1}e^{v})^k-1}{2^{-1}e^v-1 } \leq cqe^{x_0+kv}\ ] ] where the last inequality requires @xmath243 . finally , applying lemma [ l : approx1 ] on @xmath244 $ ] , @xmath245}^\infty l_x^2 } \leq c(qe^{x_0+kv}+q{\textnormal{sech}}(x_0+t_1v ) ) \leq cq e^{-v^{1-\delta}}\ ] ] as a consequence , follows .
we shall need a lemma stating that the free nonlinear flow is approximated by the free linear flow , and that the perturbed nonlinear flow is approximated by the perturbed linear flow .
both estimates are consequences of the corresponding strichartz estimates ( proposition [ p : str ] ) .
crucially , the hypotheses and estimates of this lemma depend only on the @xmath0 norm of the initial data @xmath246 .
below , is applied with @xmath247 , and @xmath248 is independent of @xmath30 ; thus @xmath30 does not enter adversely into the analysis .
[ l : approx ] if @xmath249 and @xmath250 such that @xmath251 , then @xmath252}^\infty l_x^2 } \leq c_2t_b^{1/2}\| \varphi\|_{l^2}^3\ ] ] @xmath253}^\infty l_x^2 } \leq c_2t_b^{1/2 } \| \varphi\|_{l^2}^3\ ] ] where @xmath208 and @xmath209 depend only on constants appearing in the strichartz estimates .
in particular , they are independent of @xmath29 .
estimate is in fact a special case of obtained by taking @xmath254 .
let @xmath255 so that @xmath256 with @xmath257 .
let us define @xmath258}^\infty l_x^2 \cap l_{[0,t_b]}^6l_x^6 \,,\ ] ] with the natural norm , @xmath259 .
we apply proposition [ p : str ] with , in the notation of that proposition , @xmath260 , @xmath261 , @xmath262 , @xmath263 , and then again with @xmath264 , @xmath265 , @xmath263 , to obtain @xmath266}^{6/5}l_x^{6/5 } } \,.\ ] ] the generalized hlder inequality , @xmath267 applied with @xmath268 , @xmath269 and @xmath270 , @xmath271 , gives @xmath272}^6l_x^6}^2 \|h\|_{l^2_{[0,t_b ] } l_x^2 } \\ & \leq c
t_b^{1/2}\|h\|_{l_{[0,t_b]}^6l_x^6}^2 \|h\|_{l_{[0,t_b]}^\infty l_x^2 } \\ & \leq c t_b^{1/2}\|h\|_x^3\ ,
. \end{split}\ ] ] another application of the homogeneous strichartz estimate shows that @xmath273 and consequently , @xmath274 by continuity of @xmath275 in @xmath224 , if @xmath276 , @xmath277 substituting into yields the result .
now we proceed to apply lemma [ l : approx ] .
set @xmath278 , and apply on @xmath190 $ ] to obtain @xmath279 \\ & = e^{-i(t - t_1)h_q}[u(\cdot , t_1 ) ] + \mathcal o(v^{-\delta/2})\end{aligned}\ ] ] by combining this with , @xmath280 + \mathcal o(v^{-\delta/2})\ ] ] by proposition [ p : as ] with @xmath281 for @xmath282 and @xmath283 for @xmath284 , @xmath285 , and @xmath286 replaced by @xmath287 , @xmath288(x ) \\ & = \begin{aligned}[t ] & t ( v ) e^{-i(t_2-t_1)h_0}[e^{ixv}{\textnormal{sech}}(x - x_0-vt_1)](x ) \\
& + r ( v ) e^{-i(t_2-t_1)h_0}[e^{-ixv}{\textnormal{sech}}(x+x_0+vt_1)](x ) \\ & + { \mathcal o } ( v^{-1 } ) \end{aligned } \end{aligned}\ ] ] by combining , and , @xmath289 & t(v)e^{-it_1v^2/2}e^{it_1/2}{\textnormal{nls}_0}(t_2-t_1)[e^{ixv}{\textnormal{sech}}(x - x_0-vt_1)](x ) \\ &
+ r(v)e^{-it_1v^2/2}e^{it_1/2}{\textnormal{nls}_0}(t_2-t_1)[e^{-ixv}{\textnormal{sech}}(x+x_0+vt_1)](x)\\ & + \mathcal{o}(v^{-\delta/2 } ) \end{aligned}\ ] ] by noting that @xmath290 \\ & = e^{-i(t_2-t_1)v^2/2}e^{i(t_2-t_1)/2}e^{ixv}{\textnormal{sech}}(x - x_0-t_2v)\end{aligned}\ ] ] and @xmath291 \\
& = e^{-i(t_2-t_1)v^2/2}e^{i(t_2-t_1)/2}e^{-ixv}{\textnormal{sech}}(x+x_0+t_2v)\end{aligned}\ ] ] we obtain .
let @xmath292 .
label @xmath293(x - x_0-tv)\ ] ] for the transmitted ( right - traveling ) component and @xmath294(x+x_0+tv)\ ] ] for the reflected ( left - traveling ) component . by appendix [ localization ] , for each @xmath295 , there is a constant @xmath296 and an exponent @xmath297 such that @xmath298 uniformly on the time interval @xmath192 $ ] .
we shall need the following perturbation lemma , again a consequence of the strichartz estimates .
[ l : approx6 ] let @xmath299 .
if @xmath204 , @xmath205 , and @xmath300 then @xmath301}^\infty l_x^2 } & \leq c_2\left(\|w(\cdot , t_a)\|_{l_x^2 } + \frac{c(k){\langle}q{\rangle}(\log v)^{\sigma(k)}}{v^{k(1-\delta)}}\right)\end{aligned}\ ] ] the constants @xmath208 , @xmath209 depend only on constants appearing in the strichartz estimates and are in particular independent of @xmath29 and @xmath30 .
we write the equation satisfied by @xmath210 : @xmath302 & -|w+u_{\textnormal{tr}}+u_{\textnormal{ref}}|^2(w+u_{\textnormal{tr}}+u_{\textnormal{ref } } ) + |u_{\textnormal{tr}}|^2u_{\textnormal{tr}}+ |u_{\textnormal{ref}}|^2u_{\textnormal{ref}}\\ & + q\delta_0(x)u_{\textnormal{tr}}-q\delta_0(x)u_{\textnormal{ref}}\\ \end{aligned}\\ & = \begin{aligned}[t ] & -|w|^2w - ( 2(u_{\textnormal{tr}}+u_{\textnormal{ref}})|w|^2 + ( \bar u_{\textnormal{tr}}+ \bar u_{\textnormal{ref}})w^2 ) - ( 2|u_{\textnormal{tr}}+u_{\textnormal{ref}}|^2w + ( u_{\textnormal{tr}}+u_{\textnormal{ref}})^2\bar w)\\ & -\underbrace{(u_{\textnormal{tr}}^2\bar u_{\textnormal{ref}}+ 2u_{\textnormal{ref}}|u_{\textnormal{tr}}|^2 + u_{\textnormal{ref}}^2\bar u_{\textnormal{tr}}+ 2u_{\textnormal{tr}}|u_{\textnormal{ref}}|^2)}_{{\textnormal{tr}}-{\textnormal{ref}}\textnormal { interaction}}+\underbrace{q\delta_0(x)u_{\textnormal{tr}}}_{{\textnormal{tr}}-\textnormal{delta } } -\underbrace{q\delta_0(x)u_{\textnormal{ref}}}_{{\textnormal{ref}}-\textnormal{delta } } \\ \end{aligned}\end{aligned}\ ] ] @xmath210 is estimated using proposition [ p : str ] and proposition [ p : str ] . the cubic , quadratic , and linear in @xmath210 terms on the first line are estimated exactly as was done in the proof of lemma [ l : approx1 ] . for the `` @xmath303 interaction terms '' ( taking @xmath304 as a representative example ) , we apply proposition [ p : str ] with @xmath305 , @xmath306 and estimate as @xmath307}^{4/3}l_x^1 } \leq c(t_b - t_a)^{3/4}\|u_{\textnormal{tr}}\|_{l_{[t_a , t_b]}^\infty l_x^2}\|u_{\textnormal{ref}}u_{\textnormal{tr}}\|_{l_{[t_a , t_b]}^\infty l_x^2}\ ] ] @xmath308 by mass conservation for the free nonlinear flow , and @xmath309}^\infty l_x^2 } & \leq \|u_{\textnormal{tr}}u_{\textnormal{ref}}\|_{l_{[t_a , t_b]}^\infty l_{x<0}^2 } + \|u_{\textnormal{tr}}u_{\textnormal{ref}}\|_{l_{[t_a , t_b]}^\infty l_{x>0}^2}\\ & \leq \|u_{\textnormal{ref}}\|_{l_{[t_a , t_b]}^\infty l_x^\infty } \|u_{\textnormal{tr}}\|_{l_{[t_a , t_b]}^\infty l_{x<0}^2 } + \|u_{\textnormal{tr}}\|_{l_{[t_a , t_b]}^\infty l_x^\infty } \|u_{\textnormal{ref}}\|_{l_{[t_a , t_b]}^\infty l_{x>0}^2}\end{aligned}\ ] ] now @xmath310(x)\|_{l_t^\infty l_x^\infty } \\ & \leq \|{\textnormal{nls}_0}(t)[r(v){\textnormal{sech}}](x)\|_{l_t^\infty l_x^2}^{1/2 } \|\partial_x { \textnormal{nls}_0}(t)[r(v){\textnormal{sech}}](x)\|_{l_t^\infty l_x^2}^{1/2 } \\ & \leq c\end{aligned}\ ] ] by mass and energy conservation of the free nonlinear flow . similarly , @xmath311 . by this and , the above yields @xmath312}^\infty l_x^2 } \leq \frac { c(k)(\log v)^{\sigma(k)}}{v^{k(1-\delta)}}\ ] ] thus , by , @xmath313}^{4/3}l_x^1 } \leq \frac{c(k)(\log v)^{\sigma(k)}}{v^{k(1-\delta)}}\ ] ] and similarly for all other `` @xmath303 interaction '' terms .
now we address the `` @xmath314delta '' and `` @xmath315delta '' terms ( working with @xmath316 as the representative of both ) . by proposition [ p :
str ] , we estimate as @xmath317}^{4/3 } } \leq c(t_b - t_a)^{3/4}q \|u_{\textnormal{tr}}(0,t)\|_{l_{[t_a , t_b]}^\infty}\ ] ] by , @xmath318}^{4/3 } } \leq c(k)q(t_b - t_a)^{3/4 } \frac{(\log v)^{\sigma(k)}}{v^{k(1-\delta)}}\ ] ] collecting , , and the estimates for cubic , quadratic , and linear terms in @xmath210 ( as exposed in lemma [ l : approx1 ] ) , we have , with @xmath219}^\infty l_x^2 } + \|w\|_{l_{[t_a , t_b]}^6 l_x^6}$ ] @xmath319 if @xmath320 , then the first - order @xmath210-term on the right side can be absorbed by the left , giving @xmath321 by continuity of @xmath223 in @xmath224 , if @xmath322 we have @xmath323 completing the proof .
assume that @xmath324 has been fixed .
choose @xmath325 large so that @xmath326 .
then the coefficient appearing in lemma [ l : approx6 ] is bounded by @xmath327 now take @xmath30 sufficiently large in terms of @xmath328 and @xmath147 ( thus in terms of @xmath18 ) so that the above is bounded by @xmath329 .
now we implement lemma [ l : approx6 ] . for convenience of exposition , we take @xmath227 , @xmath228 .
let @xmath330 be the integer such that @xmath331 .
we then apply lemma [ l : approx6 ] successively on the intervals @xmath332 , \ldots , [ t_2+\ell-1,t_2+\ell]$ ] as follows . applying lemma [ l : approx6 ] on @xmath332 $ ]
, we obtain @xmath333}^\infty l_x^2 } \leq 2(\|w(\cdot , t_2)\|_{l_x^2 } + v^{-1})\ ] ] applying lemma [ l : approx6 ] on @xmath334 $ ] and combining with the above estimate , @xmath335}^\infty l_x^2 } \leq 2 ^ 2\|w(\cdot , t_2)\|_{l_x^2 } + ( 2 ^ 2 + 2)v^{-1}\ ] ] continuing up to the @xmath330-th step and then collecting all of the above estimates,@xmath336}^\infty l_x^2 } \leq 2^\ell\|w(\cdot , t_2)\|_{l_x^2 } + ( 2^\ell + \cdots + 2)v^{-1}\ ] ] since @xmath337 and @xmath338 , @xmath339}^\infty l_x^2 } \leq cv^{1-\frac{3}{2}\delta}\ ] ] thus proving .
now we complete the proof of the main theorem and obtain . by and
, @xmath340 since @xmath341 , implies @xmath342 , which combined with gives and proves theorem [ th:1 ] .
in this section , we prove theorem [ th:2 ] . we note that the proof of theorem [ th:1 ] presented in [ proof ] in fact provided a more complete long - time description of the solution : @xmath343 & e^{-itv^2/2}e^{it_2/2}e^{ixv}{\textnormal{nls}_0}(t - t_2)[t(v){\textnormal{sech}}](x - x_0-tv ) \\ & + e^{-itv^2/2}e^{it_2/2}e^{-ixv}{\textnormal{nls}_0}(t - t_2)[r(v){\textnormal{sech}}](x+x_0+tv ) \\ & + \mathcal{o}_{l_x^2}(v^{1-\frac{3}{2}\delta } ) , \qquad \frac{|x_0|}{v } + v^{-\delta}\leq t \leq c(1-\delta)\log v \end{aligned}\ ] ] where @xmath344 , @xmath345 are defined in and @xmath169 denotes the solution to the nls equation @xmath346 ( without potential ) and initial data @xmath347 .
it thus suffices to obtain the resolution of @xmath348 $ ] and @xmath349 $ ] into solitons plus radiation decaying in @xmath350 . by the phase invariance of the free nonlinear flow @xmath351 = \frac{t(v)}{|t(v)|}{\textnormal{nls}_0}(t - t_2)[|t(v)|{\textnormal{sech}}]\ ] ] and similarly for @xmath349 $ ] . since @xmath352
, we apply asymptotics proved of appendix [ alphasech ] using the inverse scattering method . when @xmath353 or @xmath354 is equal to @xmath355 we use the result of @xcite recalled in .
the result obtained by these substitutions differs from that stated in theorem [ th:2 ] by a factor of @xmath356 for @xmath357 , owing to the fact that @xmath358 . but differs from @xmath19 by @xmath359 , and thus omitting it only introduces a discrepancy of @xmath360 in both @xmath176 and @xmath350 .
there is a similar inconsequential disparity in the @xmath361 part .
let @xmath362 and @xmath363 * notational conventions*. we denote @xmath364 by @xmath365 hereafter
. the @xmath366 and @xmath126 dependence of @xmath367 will be routinely dropped .
the constants @xmath368 , @xmath369 , and the polynomials @xmath370 that appear below may change ( enlarge ) from one line to the next without comment .
the constants @xmath371 depend on the fixed function @xmath372 . the solution @xmath373 satisfies conservation of mass and conservation of energy , which means that the integrals @xmath374 are independent of time @xmath375 . since @xmath376 , we have @xmath377 and it follows from the @xmath378 and @xmath379 conservation that @xmath380 , where @xmath28 depends on @xmath381 and @xmath382 .
in fact , there are an infinite number of conserved integrals , @xmath383 , with integrands defined inductively as follows @xmath384 see @xcite for a proof of this fact ( rescaling time and putting @xmath385 produces an agreement with our slightly different convention ) . the inductive definition of @xmath386 and the sobolev embedding theorem can now be used to show that , for @xmath387 , @xmath388 and hence for @xmath389 , we have @xmath390 where @xmath391 depends upon sobolev norms of the initial data @xmath246 of at most order @xmath330 .
we now elaborate on how to obtain .
an inductive argument using shows that for @xmath116 , @xmath386 is of the form @xmath392 where @xmath393 indicates a linear combination of terms of degree @xmath394 and _ cumulative _ order @xmath395 , or more precisely terms of the form @xmath396 and @xmath397 is either @xmath373 or @xmath398 . to prove for @xmath387 , one uses for @xmath399 and it only remains to verify that for any @xmath400 and @xmath401 , @xmath402 we now show this .
note that in , we may assume without loss of generality that @xmath403 .
* case 1*. @xmath404 .
it follows that @xmath405 for all @xmath406 and @xmath407 .
we estimate as : @xmath408 by sobolev embedding . * case 2*. @xmath409 . in this case , we begin by integrating by parts to obtain @xmath410 the leibniz rule expansion is @xmath411 where the sum is over @xmath412-tuples @xmath413 such that @xmath414 and @xmath415 is some constant depending on @xmath413 . by adding the @xmath416 and @xmath413 constraints
, we obtain that @xmath417 and thus there is at most one index @xmath418 ( @xmath419 ) such that @xmath420 and for all remaining @xmath421 ( @xmath422 , @xmath423 ) we have @xmath424 .
( if no such @xmath418 exists , take @xmath418 to be any fixed index @xmath425 . ) by substituting into , we estimate as @xmath426 again by sobolev embedding .
this concludes the proof of , thus , and thus . using that the commutator @xmath427=0 $ ] and some integration by parts manipulations
, we have the pseudoconformal conservation law : @xmath428 from this , for @xmath429 , and the gagliardo - nirenberg estimate @xmath377 , we have @xmath430 where @xmath28 depends on @xmath431 , @xmath381 , and @xmath432 .
we want to show that more generally , for each @xmath433 , @xmath116 , we have @xmath434 here @xmath368 is a constant depending on @xmath147 and weighted sobolev norms of the initial data @xmath246 ( up to order @xmath435 ) , and @xmath369 is a positive exponent depending upon @xmath147 .
we are not concerned with obtaining the optimal value of @xmath369 ; the mere fact that the bound in is power - like in @xmath126 , as opposed to exponential in @xmath126 , suffices for our purposes . in our proof
, both @xmath368 and @xmath369 will be increasing with @xmath147 , and will go to @xmath22 as @xmath436 .
let @xmath437 and @xmath438 .
note that both operators have the commutator property @xmath439=0\ , , \ \
j=0 , 1\,.\ ] ] we first claim that for each @xmath116 , there exists a constant @xmath296 and an exponent @xmath297 such that @xmath440 when we wish to consider a composition of the form @xmath441 and do not care to report whether each operator in the composition is @xmath442 or @xmath443 ,
we will instead write the composition as @xmath444 . we prove by induction on @xmath147 . when @xmath230 , is just the mass conservation law .
suppose that holds for @xmath445 ; we aim to prove it holds for @xmath147 .
the main ingredient ( in addition to the inductive hypothesis ) is .
fix @xmath446 , and apply the operator @xmath447 to the equation , pair with @xmath448 , integrate in @xmath366 , take twice the real part , and appeal to to obtain @xmath449 note that @xmath450 and @xmath451 both of these product rules take the form @xmath452 where @xmath453 is a polynomial in @xmath126 of degree @xmath454 .
thus we see that @xmath455 where @xmath456 is a polynomial in @xmath126 . substituting into ,
we obtain two terms : the first is zero since it is the real part of a purely imaginary number ; the second is estimated by the hlder inequality to obtain : @xmath457 by sobolev embedding estimates , , and the induction hypothesis , we have @xmath458 from which follows . now to deduce from , we just note that since @xmath459 , there are polynomials @xmath460 such that the following relation holds : @xmath461 let us now consider the application of to obtain in the phase 3 analysis .
we have @xmath462 for @xmath463 .
if @xmath119 , then @xmath464 thus @xmath465(x - x_0-tv)\|_{l_x^2 } \\ & = \|x^k { \textnormal{nls}_0}(t - t_2)[t(v){\textnormal{sech}}](x)\|_{l_x^2 } \\ &
\leq c(k){\langle}t - t_2 { \rangle}^{\sigma(k)}\end{aligned}\ ] ] by , which gives the first estimate in .
the second is obtained similarly . to obtain the third
, we note that for @xmath463,@xmath466(x - x_0-tv)|^2 \
, dx\end{aligned}\ ] ] and this can be estimated by @xmath467(x - x_0-tv ) \|_{l_{x<0}^2 } \| \partial_x { \textnormal{nls}_0}(t - t_2)[t(v){\textnormal{sech}}](x - x_0-tv ) \|_{l_{x<0}^2 } \,.\ ] ] using as before establishes @xmath468 ( x - x_0-tv ) \|_{l_{x<0}^2 } \leq c(k+j ) { \langle}t - t_2 { \rangle}^{\sigma(k+j)}\ ] ] replacing @xmath469 by @xmath368 and @xmath470 by @xmath369 , we obtain the bound .
finally , we note that the fourth bound in is similar to the third .
this appendix is devoted to showing that for @xmath472 @xmath473 a more precise understanding of error terms is possible thanks to advances in the study asymptotics for integrable nonlinear waves @xcite , @xcite . since we do not know an exact reference for we present a proof of this simpler asymptotic result .
it is based on the now classical work on the inverse scattering method initiated for nls by zakharov - shabat @xcite
see @xcite,@xcite for discussion and references . for the reader
s convenience , especially in view of different conventions used in different sources for our argument , we review all the needed aspects of the method . in the case of
@xmath474 we can use the result of @xcite to conclude that @xmath475 a slightly inaccurate statement similar to was given in @xcite and the calculation of the scattering matrix in that paper was our starting point in obtaining .
we present a quick review of this celebrated method .
thus , let us consider two operators acting on @xmath476 : @xmath477 where @xmath478 then @xmath479 = - \frac i 2 q_{xx } + iq^3\ ] ] which is checked by using @xmath480 it is now the case that @xmath481 \
\longleftrightarrow \ i\partial_t u + \frac12 \partial_x^2 u + |u|^2u=0 \,,\ ] ] and , since we are solving nls , we assume that these equivalent equations hold .
we now consider scattering theory for the time dependent operator @xmath482 .
for that we introduce special solutions to @xmath483 with prescribed asymptotic behaviour : @xmath484 \ , , \ \
\psi ( x , \lambda ) \simeq \left [ \begin{array}{l } \ 0 \\ e^{i x \lambda } \end{array } \right ] \ , , \
\ x \longrightarrow + \infty \\ & \varphi ( x , \lambda ) \simeq \left [ \begin{array}{l } e^{-i x \lambda } \\ \ 0 \end{array } \right ] \
\ \bar \varphi ( x , \lambda ) \simeq \left [ \begin{array}{l } \ 0 \\ e^{i x \lambda } \end{array } \right]\ , , \ \ x \longrightarrow - \infty \ , , \end{split}\ ] ] see instance ( * ? ? ?
* sect.i.5 ) . here for vector valued functions , @xmath485^t $ ] , if @xmath486^t $ ] .
each pair of solutions forms a basis for the solution set and , for @xmath487 , @xmath488 another consequence comes from . if @xmath489 then we see that @xmath490 and hence @xmath491 now we note that for @xmath492 , @xmath493 , as @xmath494 , and the asymptotic behaviour gives @xmath495 , @xmath496 .
more generally we conclude that @xmath497 the solutions @xmath498 and @xmath68 have analytic extensions in @xmath499 to the upper half plane and @xmath500 and @xmath501 to the lower half plane .
same is true for @xmath502 and @xmath503 respectively .
except in very special cases ( such as our potential @xmath504 ) @xmath505 does not have an analytic extension off the real axis .
the reflection coefficient is defined as @xmath506 we assume that @xmath507 has at most one zero and that it can only lie in @xmath508 , @xmath509 . that zero , @xmath510 , corresponds to an @xmath511 eigenfuction of @xmath482 , and at @xmath512 ,
the two solutions are proportional : @xmath513 the scattering data is given by the triple @xmath514 the evolution of the scattering data is easily obtained from : @xmath515 in fact , we can use to see @xmath516 independence of @xmath498 and @xmath500 shows that , remarkably , @xmath517 which gives the first part of . from we
see that , @xmath518 so that , @xmath519 that gives .
the justification of this formal calculation depends on @xmath520 and we refer to , for instance , ( * ? ? ?
* section i.7 ) for a full proof .
it is now universally acknowledged that the best way to obtain long time asymptotics for the inverse of and is by solving a riemann - hilbert problem @xcite , ( * ? ? ?
* chapter ii ) . to recall this method
let us consider the following matrix valued function of @xmath521 , depending parametrically on @xmath522 : @xmath523 , & { \mathop{\rm im}\nolimits}\lambda > 0 \\ \ & \ \\ { [ } \bar \psi ( x , \lambda ) e^ { i \lambda x } , \bar a ( \bar \lambda)^{-1 } \bar \varphi ( x , \lambda ) e^ { - i \lambda x } { ] } , & { \mathop{\rm im}\nolimits}\lambda < 0 \end{array } \right.\ ] ] the properties of @xmath498 and @xmath68 ( see for instance ( * ? ? ?
* section i.5 ) ) imply that @xmath524 , uniformly in compact sets . from we
see that @xmath525 \left [ \begin{array}{ll } \ \ \ \ \ \ \ 1 & 0 \\ r ( \lambda ) e^ { 2 i ( \lambda x + \lambda^2 t ) } & 1 \end{array } \right ] \ , , \\
& \psi ( \lambda - i 0 , x ) = [ \bar \psi ( x , \lambda ) e^ { i \lambda x } , \psi ( x , \lambda ) e^{- i \lambda x } ] \left [ \begin{array}{ll } 1 & - { \bar r ( \lambda ) } e^ { - 2 i ( \lambda x + \lambda^2 t ) } \\ 0 & \ \ \ \ \ \ \ \ 1 \end{array } \right ] \ , , \ \
\in { { \mathbb r}}\ , .
\end{split}\ ] ] hence , the boundary values of @xmath526 satisfy @xmath527 \ , .
\end{gathered}\end{gathered}\ ] ] if @xmath507 has no zeros in @xmath528 then the riemann - hilbert problem is to construct @xmath526 satisfying and .
liouville s theorem readily shows that it is unique .
if @xmath507 has a zero in @xmath529 , and in our presentation we allow at most one , @xmath510 , we have to consider a riemann - hilbert problem in which @xmath526 is allowed to have singularities at @xmath510 and @xmath530 .
the structure of that singularity can be seen in : @xmath531 \ , , \\ & { \operatorname{res } } _ { \lambda = \bar \lambda_0 } \psi = \psi ( \bar \lambda_0 ) \left [ \begin{array}{ll } 0 & e^ { - 2 i \bar{\lambda_0 } x } \bar \gamma_0 ' \\ 0 & \ \ \ \ 0 \end{array } \right ] \ , , \ \
\gamma_0 ' { \stackrel{\rm{def}}{=}}\frac{\gamma_0 } { a ' ( \lambda_0 ) } \ , .
\end{split}\ ] ] since @xmath507 can be reconstructed from @xmath532 and @xmath510 ( see for instance ( * ? ? ?
* chapter i , ( 6.23 ) ) ; in our case it will be explicit ) the riemann - hilbert problem in the case of one singularity is to find @xmath533 which in addition to and satisfies . a standard way to read off @xmath534 from @xmath535 follows from high frequency asymptotics of @xmath536
( see for instance ( * ? ? ? * ( 18 ) ) ) : @xmath537 + \frac{1 } { 2 i \lambda } \left [ \begin{array}{l } \ \ \ \ \ u ( t , x )
\\ \int_x^\infty | u ( t , y ) |^2 dy \end{array } \right ] + { \mathcal o } \left ( \frac1 { | \lambda|^2 } \right ) \,,\ ] ] so that @xmath538 we conclude this brief review by describing a reduction of the problem with prescribed singularities to a problem with @xmath533 analytic in @xmath539 . to do that we follow ( * ? ? ? * section ii.2 ) by considering a reformulation of the riemann - hilbert problem : @xmath540 \,,\ \ a_+ ( \lambda ) { \stackrel{\rm{def}}{=}}a ( \lambda ) \ , , \
\ a_- ( \lambda ) { \stackrel{\rm{def}}{=}}\bar a ( \bar \lambda ) \ , , \ \ \pm { \mathop{\rm im}\nolimits}\lambda > 0 \ , , \\
g_+ ( \lambda + i 0 ) g_- ( \lambda -i0 ) = g ( \lambda ) \ , , \ \
g ( \lambda ) { \stackrel{\rm{def}}{=}}\left [ \begin{array}{ll } \ \ \ 1 & - \bar b ( \lambda ) \\ - b ( \lambda ) & \ \ \ 1 \end{array } \right ] \ , .
\end{gathered}\ ] ] the operators @xmath541 are now analytic in @xmath542 ( in fact , @xmath543 ) but their ranks drop precisely at @xmath544 and @xmath530 respectively .
the condition becomes @xmath545 \ , , \ \
{ \operatorname{ker}}g_- ( \bar \lambda_0 ) = { \rm span}_{{\mathbb c}}\left [ \begin{array}{l } - \bar \gamma_0 e^ { - 2 i \bar \lambda_0 x } \\ \ \ \ \ 1 \end{array}\right ] \,.\ ] ] we now look for @xmath546 , analytic in @xmath547 , with @xmath548 analytic in @xmath549 , @xmath550 , and such that @xmath551 are nonsingular matrices .
we note this requires @xmath552 to have a pole at @xmath510 and @xmath553 , at @xmath530 .
that is natural since we are adding to the ranks of @xmath541 .
the condition mean that @xmath554 \ , , \ \
{ \operatorname{ker}}b ( \bar \lambda_0 ) = { \rm span}_{{\mathbb c}}\widetilde g_- ( \bar \lambda_0 ) \left [ \begin{array}{l } - \bar \gamma_0 e^ { - 2 i \bar \lambda_0 x } \\ \ \ \ \ 1 \end{array}\right ] \,.\ ] ] this determines @xmath555 uniquely as a _ blaschke - potapov factor _ : @xmath556 see ( * ? ? ?
* chapter ii , ( 2.17)-(2.27 ) ) . hence to solve the riemann - hilbert problem
, we first solve the problem @xmath557 we then define @xmath558\ , , \ \
\widetilde g_+ ( \lambda ) = \widetilde g_- ( \bar \lambda ) ^ * \ , , \ \ { \mathop{\rm im}\nolimits}\lambda < 0 \,,\ ] ] from which , using , we construct @xmath546 .
then @xmath559 and we can finally use to obtain @xmath560 .
in particular the long time behaviour of @xmath561 is determined by the longtime behaviour of @xmath562 .
the basic structure of the long time behaviour of @xmath563 solving can be obtained from the _ manakov ansatz _ for the solution of see ( * ? ? ?
* section 2 ) and references given there . to describe it we define
@xmath564 where @xmath565 , @xmath566 , and @xmath567 solves a scalar riemann - hilbert problem , @xmath568 see ( * ? ? ?
* proposition 2.12 ) for a detailed list of properties of @xmath569 ( stated in the defocusing case @xmath570 to see the properties of @xmath571 we use the following elementary lemma : using this lemma one checks that @xmath579 if we defined @xmath580 by putting @xmath581 in this shows that @xmath582 , and in fact a more precise statement can be obtained by using the second component in
. we can now use to obtain an approximation to @xmath546 as @xmath583 : @xmath584 where @xmath585 is as in with @xmath586 replaced by @xmath587 hence to obtain a long time approximation for a solution we apply the procedure of ( * ? ? ?
* section ii.5 ) to @xmath588 since that corresponds to using with @xmath589 this gives we now apply lemma [ lb:2 ] to obtain .
for that we need to find the scattering data for @xmath597 . that is done by a well known computation @xcite , ( *
* sect.3.4 ) which reappears in many scattering theories , from the free @xmath598-matrix in automorphic scattering , to eckhardt barriers in quantum chemistry .
we quote the results : @xmath599 we note that in this special case @xmath600 and @xmath601 are meromorphic in @xmath602 ( with infinitely many `` nonphysical '' poles ) .
also , @xmath603 we need to compute @xmath604 and @xmath605 . in general when @xmath606 is even then @xmath607 by symmetry considerations . here
we see it by using ( * ? ? ?
* chapter ii , ( 2.6 ) ) which shows that @xmath608 thus the formula in lemma [ lb:2 ] results in @xmath607 . to compute @xmath605 we need to find the following integral @xmath609 p.a
deift , a.r .
its , and x. zhou , _ long - time asymptotics for integrable nonlinear wave equations _ , in _ important developments in soliton theory _
, 181204 , springer ser .
nonlinear dynam . , springer , berlin , 1993 . | we study the gross - pitaevskii equation with a repulsive delta function potential .
we show that a high velocity incoming soliton is split into a transmitted component and a reflected component .
the transmitted mass ( @xmath0 norm squared ) is shown to be in good agreement with the quantum transmission rate of the delta function potential .
we further show that the transmitted and reflected components resolve into solitons plus dispersive radiation , and quantify the mass and phase of these solitons . |
the origin of neutrino mass is one of the unsolved issue in the standard model ( sm ) .
this mystery could be solved by a higgs triplet extension of the sm called type - ii seesaw model @xcite .
the novel feature of a type - ii seesaw model is the existence of a doubly charged higgs @xmath5 in the higgs triplet @xmath6 . in the original type - ii seesaw model
, @xmath5 decays into two same - sign charged leptons @xmath7 or w bosons @xmath8 according to the triplet vacuum expectation value ( vev ) when masses in the multiplet are degenerate .
thus the experimental searches of doubly charged higgs bosons have focused on the signals of same - sign dilepton or w bosons @xcite . the extension of type - ii seesaw model by including second higgs doublet can lead new effect in doubly charged higgs production and decays @xcite .
the most attractive new effect is the dimension-3 terms in scalar potential , read by @xmath9 ( j , k = 1,2 ; @xmath10=1,2,3 ) , where @xmath11 is a higgs doublet . since the coefficients @xmath12 can be of order of electroweak ( ew ) scale , the new terms lead a large mixing angle between the singly charged higgs of doublet ( @xmath13 ) and of triplet ( @xmath14 ) .
consequently the doubly charged higgs can be produced via qcd process owing to the the mixing effect . in this study
we focus on the new production and decay patterns of @xmath5 .
we then carry out a simulation study for the signal of @xmath5 at the lhc and estimate the significance of discovering the doubly charged higgs .
to investigate the production and decay of doubly charged higgs boson @xmath5 , we first discuss the relevant interactions of charged higgs bosons in the two higgs doublet(thd)-type - ii seesaw model .
the gauge interactions and yukawa couplings of @xmath6 and @xmath15 are same as original type - ii seesaw model and thd model ( type - ii ) respectively .
we thus do not explicitly write down them here ; more details are given in refs @xcite .
the higgs potential is written such that @xmath16 \
, , \nonumber \\
v_\delta & = m^2_\delta tr \delta^\dagger \delta + \lambda_{9 } ( tr \delta^\dagger \delta)^2 + \lambda_{10 } tr ( \delta^\dagger \delta)^2\ , , \nonumber \\ v_{h_1h_2\delta } & = \left ( \mu_1 h^t_1 i\tau_2 \delta^{\dagger } h_1 + \mu_2 h^t_2 i \tau_2 \delta^\dagger h_2 + \mu_3 h^t_1 i\tau_2 \delta^\dagger h_2 + h.c .
\right ) \nonumber \\ & + ( \lambda_6 h^\dagger_1 h_1 + \bar \lambda_6 h^\dagger_2 h_2 ) tr \delta^\dagger \delta + h^\dagger_1 ( \lambda_7 \delta \delta^\dagger + \lambda_8 \delta^\dagger
\delta ) h_1 + h^\dagger_2 \left ( \bar \lambda_7 \delta \delta^\dagger + \bar \lambda_8 \delta^\dagger \delta \right ) h_2\ , , \label{eq : v } \end{aligned}\ ] ] where @xmath17 and @xmath18 denote the scalar potential of thd and of pure triplet , and @xmath19 is the part involving @xmath20 , @xmath21 and @xmath6 .
the vev of the triplet @xmath22 is required to be @xmath23 where @xmath24 is the vev of @xmath15 .
we only keep the leading power for @xmath22 in eq .
( [ eq : v_d ] ) since it should satisfy @xmath25 from @xmath26 parameter measurement .
interestingly , the parameters @xmath27 as large as electroweak scale can provide @xmath28 if the parameters satisfy the relation @xmath29 in contrast to original type - ii seesaw model . in our analysis
, we adopt the relation for simplicity .
two physical singly charged higgs bosons are provided by both thd sector and higgs triplet which are denoted by @xmath13 and @xmath30 respectively .
the mass eigenstates are obtained as a combination of @xmath13 and @xmath30 as @xmath31 the masses and their mixing angles are written by @xmath32^{1/2}\ , , \quad \tan2\theta_\pm = - \frac{2 m^2_{h^- \delta^+}}{m^2_{\delta^\pm } - m^2_{h^\pm}}\ , , \label{eq : mass_mixing } \end{aligned}\ ] ] where @xmath33 is identified as the lighter charged higgs , @xmath34 , @xmath35/(2 \sin
\beta \cos \beta ) $ ] , and @xmath36 .
the magnitude of the mixing angle @xmath37 depends on the massive parameter @xmath38 .
we are going to explore the influence of a large @xmath37 on the search of doubly charged higgs . in our model ,
two singly charged higgses can interact with quarks owing to the large mixing effect .
the @xmath39-quark interactions are given by @xmath40 ( \cos\theta_\pm h^+_1 - \sin\theta_\pm h^+_2 ) + h.c .
\ , , \label{eq : hff } \end{aligned}\ ] ] where we suppress all flavor indices , @xmath41 and @xmath42 denote the up and down type quarks , @xmath43 is the cabibbo - kobayashi - maskawa ( ckm ) matrix , @xmath44 is the diagonalized mass matrix of down ( up ) type quarks , and @xmath45 .
the doubly charged higgs could be produced by ew interactions via the s channel processes : @xmath46 except for the new mixing effect @xmath47 , the production channels are similar to those in the original type - ii seesaw model . moreover , with the sizable mixing angle @xmath47 , the on - shell @xmath5 could be produced through qcd interactions : @xmath48 the on - shell @xmath5 in eq .
( [ eq : qcdi ] ) and ( [ eq : qcdii ] ) is generated by the decay @xmath49 since we adopt mass relation of @xmath50 .
the production of @xmath2 through lighter charged higgs @xmath51 is off - shell effects and small , we therefore ignore its contributions .
for the processes in eq .
( [ eq : qcdii ] ) , the main qcd reaction is associated with the interactions of b - quark and gluons , e.g. @xmath52 . in our analysis
, we apply following conditions for the parameters in the model : @xmath53 which give sizable mixing angle @xmath47 and the mass relation of @xmath50 .
the masses of singly charged higgs and mixing angle are then obtained from eq .
( [ eq : mass_mixing ] ) .
we then calculate the cross sections of @xmath5 production processes using _
calchep 3.6.15 _ code @xcite by implementing the model .
applying the settings of eq .
( [ eq : setting ] ) and @xmath54 , we show the production cross sections for the processes in eqs .
( [ eq : ew1])-([eq : qcdii ] ) as a function of @xmath55 in fig .
[ fig : xs ] , where the collision energy at lhc is 13 tev and cteq6l pdf is used ; the dotted , dash - dotted and dash - dot - dotted lines denote the ew processes while the solid and dashed lines stand for qcd processes , respectively .
we therefore find that qcd production cross sections are significantly larger than that of ew production in our setting .
the produced @xmath5 can decay into @xmath56 , @xmath8 , @xmath57 , @xmath58 , etc . with our parameter setting ,
the dominant decay modes are @xmath59 where @xmath60 is off - shell , since we assume @xmath28 and small yukawa couplings for triplet suppressing @xmath8 and @xmath61 modes .
then the lighter singly charged higgs @xmath62 dominantly decays into @xmath63 ; the off - shell @xmath64 also coverts to @xmath63 . ) and @xmath65 are used .
( reproduced from @xcite.),width=264 ]
in this section , we discuss the simulation study of possible signal and background events .
we then estimate the significance of discovering the doubly charged higgs after applying relevant kinematic cuts ( see also @xcite for details ) . in order to generate the simulation events
, we employ the madgraph / madevent5 @xcite and use pythia6 @xcite to deal with the fragmentation of hadronic effects , the initial - state radiation ( isr ) and final - state radiation ( fsr ) effects , and the decays of sm particles e.g. @xmath66-boson , @xmath67-quark , etc .
in addition , the generated events are also run though the pgs4 detector simulation @xcite .
the @xmath5 is dominantly produced by the qcd production processes eqs .
( [ eq : qcdi ] ) and ( [ eq : qcdii ] ) , and decays as @xmath68 .
we thus require the signal events as @xmath69 for background processes we consider following final states at the lhc : @xmath70 ( + isr / fsr ) , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 and @xmath76 , where the number of jets @xmath77 for vv backgrounds is taken as @xmath78 .
@xmath79 events in @xmath80 background have been included in ew and qcd background , therefore they should be excluded . for enhancing the significance of @xmath2 signals by reducing the possible backgrounds , we need to propose some strategies of kinematical cuts .
for excluding the soft leptons and jets , when we generate the events by event generator , we set the preselection conditions for leptons and jets to be @xmath81 where @xmath82 is the transverse momentum and @xmath83 is pseudo - rapidity with @xmath84 being the scattering angle in the laboratory frame .
furthermore we apply the selection cuts @xmath85 where @xmath86 denotes the number of b - jet , @xmath87 stands for the second highest @xmath82 charged lepton .
finally we estimate significance which is defined as @xcite @xmath88}$ ] , where @xmath89 and @xmath90 denote the number of signal and background events , respectively .
the left ( right ) panel of fig .
[ fig : significance ] is the estimated significance ( luminosity to get 5@xmath91 discovery ) as a function of @xmath55 . by the figure
, one can find that the doubly charged higgs with a mass lower than 330 gev can be discovered at the lhc with an integrated luminosity of 40 fb@xmath1 .
additionally , the doubly charged higgs with a mass of 450 gev can be discovered at the lhc with an integrated luminosity of 300 fb@xmath1 .
( left ) and luminosity for @xmath92 ( right ) as a function of @xmath55 .
the collision energy of 13 tev is applied for both plots .
( reproduced from @xcite . )
[ fig : significance],title="fig : " ] ( left ) and luminosity for @xmath92 ( right ) as a function of @xmath55 . the collision energy of 13 tev is applied for both plots .
( reproduced from @xcite . )
[ fig : significance],title="fig : " ]
we have discussed the type - ii seesaw model with two higgs doublet .
the new interaction in the model can lead the significant mixing effect between singly charged higgs of the higgs doublet and triplet .
the mixing results in new production processes and decay patterns of the doubly charged higgs . we have shown that qcd processes are the predominant effects to produce the @xmath2 , read as @xmath93 and @xmath94 owing to the mixing effect , while other higgs triplet models are arisen from ew processes .
subsequently the doubly charged higgs decays as @xmath68 .
we then have investigated the significance of doubly charged higgs signal by numerical simulation including relevant kinematical cuts .
we find that with luminosity of 40 fb@xmath1 and collision energy of 13 tev , @xmath2 with mass below @xmath3 gev could be observed at the @xmath4 level .
additionally , the observed mass of @xmath2 could be up to 450 gev when the luminosity approaches 300 fb@xmath1 .
this work is supported by the ministry of science and technology of r.o.c . under grant # : most-103 - 2811-m-006 - 030 .
we also thank the national center for theoretical sciences ( ncts ) for supporting the useful facilities .
99 m. magg and c. wetterich , phys .
b * 94 * , 61 ( 1980 ) ; g. lazarides , q. shafi and c. wetterich , nucl .
b * 181 * , 287 ( 1981 ) ; r. n. mohapatra and g. senjanovic , phys .
d * 23 * , 165 ( 1981 ) ; e. ma and u. sarkar , phys . rev .
lett . * 80 * , 5716 ( 1998 ) [ hep - ph/9802445 ] .
w. konetschny and w. kummer , phys .
b * 70 * , 433 ( 1977 ) ; j. schechter and j. w. f. valle , phys .
d * 22 * , 2227 ( 1980 ) ; t. p. cheng and l. -f .
li , phys .
d * 22 * , 2860 ( 1980 ) ; s. m. bilenky , j. hosek and s. t. petcov , phys .
b * 94 * , 495 ( 1980 ) .
s. chatrchyan _ et al . _
[ cms collaboration ] , eur .
j. c * 72 * , 2189 ( 2012 ) [ arxiv:1207.2666 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas collaboration ] , eur .
j. c * 72 * , 2244 ( 2012 ) [ arxiv:1210.5070 [ hep - ex ] ] .
v. khachatryan _ et al . _
[ cms collaboration ] , arxiv:1410.6315 [ hep - ex ] . c. h. chen and t. nomura , phys .
d * 90 * , 075008 ( 2014 ) [ arxiv:1406.6814 [ hep - ph ] ] . c. h. chen and t. nomura , phys .
d * 91 * , no . 3 , 035023 ( 2015 ) [ arxiv:1411.6412 [ hep - ph ] ] .
j. alwall , r. frederix , s. frixione , v. hirschi , f. maltoni , o. mattelaer , h .- s .
shao and t. stelzer _ et al .
_ , jhep * 1407 * , 079 ( 2014 ) [ arxiv:1405.0301 [ hep - ph ] ] . t. sjostrand , s. mrenna , p. z. skands , jhep * 0605 * , 026 ( 2006 ) . http://www.physics.ucdavis.edu/conway/research/software/pgs/pgs4-general.htm . | a study of searching for doubly charged higgs @xmath0 is performed in two - higgs - doublet extension of the type - ii seesaw model .
we find that a significant mixing effect between singly charged higgs of higgs doublet and of triplet is arisen from the scalar potential .
the mixing leads to new production processes and decay patterns of doubly charged higgs . with luminosity of 40 fb@xmath1 and collision energy of 13 tev at the lhc , we show that @xmath2 with mass below @xmath3 gev could be observed at the @xmath4 level .
moreover , for the luminosity of 300 fb@xmath1 , the observed mass of @xmath2 could reach up to 450 gev . |
axps are a small group of neutron stars ( nss ) which stand apart from other known classes of x - ray sources .
at the moment there are 7 confirmed axps plus 2 candidates .
these x ray pulsars share many peculiarities : they are all ( but one ) radio - quiet ( camilo et al . 2006 ; burgay et al .
this volume ) , exhibit x - ray pulsations with spin periods in a small range of values ( @xmath0512s ) , they have a large spin - down rate ( @xmath1ss@xmath2 , a rather high x - ray luminosity ( @xmath3 ) , and faint ir counterparts with @xmath4magnitudes ( for a recent review see woods & thompson 2004 and kaspi 2006 in this volume ) .
the nature of their x - ray emission was intriguing all along .
in fact , it is too high to be produced by the loss of rotational energy alone , but on the other hand , no hints for a companion star were found , neither through deep observations at other wavelength , nor timing the x - ray pulsations with the hope of finding doppler shifts ( israel et al . 2003a ;
mereghetti , israel & stella 1998 ) .
[ tab:1 ] lccccccc & asca1996 & sax1999 & sax2001 & chandra2002 & xmm2003 & chandra2004 & swift2005 + kt ( kev ) & 0.41@xmath5 & 0.465@xmath6 & 0.424@xmath7 & 0.475@xmath8 & 0.456@xmath9 & 0.43@xmath5 & 0.430@xmath10 + gamma & 2.51@xmath11 & 2.65@xmath12 & 2.45@xmath13 & 2.47@xmath14 & 2.792@xmath15 & 2.77@xmath16 & 2.62@xmath17 + flux & 1.5@xmath18 & 1.23@xmath19 & 1.30@xmath20 & 1.06@xmath21 & 0.87@xmath22 & 1.30@xmath23 & 1.43@xmath24 + pl flux ( % ) & @xmath25 & @xmath26 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 + @xmath32 ( d.o.f . ) & 1.05 ( 71 ) & 1.07 ( 148 ) & 1.19 ( 215 ) & 0.93 ( 430 ) & 1.14 ( 221 ) & 0.95 ( 147 ) & 1.11 ( 182 ) + at present , the model which is most successful in explaining the peculiar observational properties of axps is the `` magnetar '' model . in this scenario axps
are thought to be isolated nss endowed with ultra - high magnetic fields ( @xmath33gauss ) and their x - ray emission powered by magnetic field decay ( duncan & thompson 1992 ; thompson & duncan 1993 , 1996 ) .
this idea is strongly supported by the estimate of axps magnetic field through the classical dipole braking formula , @xmath34 gauss , which gives in all cases values above the electron critical magnetic field ( @xmath35 gauss ) .
alternative scenarios , invoking accretion from a fossil disk remnant of the supernova explosion ( van paradijs , taam & van den heuvel 1995 ; chatterjee , hernquist & narayan 2000 ; alpar 2001 ) , are still open possibilities although encounter increasing difficulties in explaining the data .
was discovered with rosat ( voges et al . 1996 ) and later on a @xmath011s modulation was found in its x - ray flux with asca ( sugizaki et al .
early measurements suggested that it was a fairly stable rotator with a spin period derivative of @xmath36ss@xmath37 ( israel et al .
however , in the last four years the source experienced two glitches , with different post - glitch recoveries ( kaspi , lackey & chakrabarty 2000 , dallosso et al .
2003 , kaspi & gavriil 2003 ) .
searches for optical / ir counterparts ruled out the presence of a massive companion ( israel et al .
very recently , two different objects were proposed by different groups , as being ir counterpart , and there still is an open debate on which one is the axp counterpart ( israel et al . 2003b ; safi - harb & west 2005 ; durant & van kerkwijk 2006 ) .
a diffuse ( @xmath38 ) radio emission at 1.4 ghz was recently reported , possibly associated with the supernova remnant g346.50.1 ( gaensler et al .
2000 ) . pulse phase spectroscopy analysis of two _ beppo_sax observations of ( israel et al . 2001 ; rea et al .
2003 ) revealed i ) a large spectral variability with the spin - phase , ii ) a strong energy dependence of the pulse profile shape , and iii ) shifts in the pulse phase between the low and the high energy profiles .
high variability of the pulse shape with energy is now detected at higher energies , up to @xmath0220 kev ( kuiper et al .
2006 ) . by analysing a _ beppo_saxobservation
taken in 2001 ( the longest pointing ever performed on this source ) , rea et al .
( 2003 ) reported the presence of an absorption line at @xmath39 kev in a phase - resolved spectrum . interpreting the feature as a cyclotron line due to resonant scattering yields a neutron star magnetic field of either @xmath40 g or @xmath41 g , in the case of electron or proton scattering , respectively . in [ observations ] we report on new xmm - newton , chandra and swift observations which were used together with the previous rosat , asca and _
beppo_sax observations to monitor the x - ray spectrum and flux of the axp
. then we report on deep infrared vlt - naco observations of this axp . in [ cyclo ]
we carefully re - address the calculation of the significance of the absorption line found around 8kev during a long _ beppo_sax observation ( rea et al .
2003 ) , we then summarise and discuss all the results in [ diss ] .
in this section we report on the observations and the data analysis of the four new x ray and the ir observations .
the results on the x ray timing analysis are reported below in the text while the x ray spectral parameters may be found in tab.1 .
all the x - ray spectra were fit by an absorbed blackbody plus a power law component ( but see also rea , zane , lyutikov & turolla in this volume for a different spectral modelling )
. giving the very high statistics we have in the xmm
newton observation , and assuming the absorption does not vary along the line of sight , we fixed for all the spectra the absorption at the xmm newton value ( phabs xspec model : @xmath42@xmath43 ; abundances from anders & grevesse 1989 ) was observed with xmm
newton between 2003 august 28th and 29th , for @xmath44 ks .
the mos cameras were operated in prime partial window mode , while the pn camera was in prime small window mode , all with the medium optical photons blocking filter .
since a higher background affected the last @xmath010ks of the observation , we used only the data during intervals in which the count rate above 10 kev was less than 0.35 counts s@xmath37 .
the source events and spectra were extracted within a circular region of 27@xmath45 centred on the peak of the point spread function of the source .
this non standard radius was used because the source was located near the edge of the chip .
the background was obtained from a source - free region of 27@xmath45 . in order to determine the spin period of we barycentered the events arrival times and obtained , through a phasefitting technique , a best spin period of @xmath46s ( all errors are at the 90% confidence level ) .
we found that the pulsed fraction of the x - ray signal ( defined as the amplitude of the best - fitting sine wave divided by the , background corrected , constant level of the emission ) is energy - dependent , and it varies from @xmath47% in the 0.52.0kev range to @xmath48% in the 6.010.0kev range .
these values are consistent with those reported for the pre - glitches _ beppo_saxobservation ( israel et al .
2001 ) while both are larger than those reported for the post - glitches _ beppo_sax observation ( rea et al . 2003 ) .
detailed results for this observation are reported in rea et al .
. band of the field of .
the chandra @xmath49 90% error circle is over - plotted ( israel et al .
we marked all the faint sources being consistent at 3@xmath50 with the axp position .
note that our astrometry ( following israel et al .
2003b ) is slightly different from the one reported by durant & van kerkwijk ( 2006 ) , this is due to a different catalogue used for the astrometry.,scaledwidth=35.0%,scaledwidth=35.0% ] was observed by the _ chandra _
advanced ccd imaging spectrometer ( acis ) , first for @xmath030ks with the high energy transmission grating spectrometer ( hetgs ) on 2002 september 9th , then for @xmath030ks in continuous clocking ( cc ) mode on 2004 july 3 . for a more detailed description of the instruments and on the data processing we defer to the _ chandra _ x - ray center ( cxc ) documents .
detailed results for these two observations are reported in rea et al .
( 2005a ) and campana et al .
( 2006 ) .
the high energy transmission grating spectrometer ( hetgs ) employs two sets of transmission gratings : the medium energy gratings ( megs ) with a range of 2.531 ( 0.45.0 kev ) and the high energy gratings ( hegs ) with a range of 1.215 ( 0.810.0 kev ) .
the hetgs spectra were imaged by acis - s , an array of 6 ccd detectors normally read - out every 3.2s .
the hetgs / acis - s combination provides an undispersed ( zeroth order ) image and dispersed spectra from the gratings .
the various orders overlap and are sorted using the intrinsic energy resolution of the acis ccds : @xmath51 0.012 for the heg and 0.023 for the meg .
the meg and heg first order count rate were only 0.5 and 0.2 cts s@xmath37 , we therefore did not expect the dispersed spectrum to be affected by pileup , while the zeroth - order image was not used in our spectral analysis because highly affected by photon pileup .
we used the standard ciao tools to create detector response files for the meg and heg @xmath52 and @xmath53 order spectra .
these were combined when the @xmath54/@xmath55 order spectra were added for the heg and meg separately .
we binned the data at 0.08 with a minimum of 30 counts per bin . to look for high - resolution spectral features ,
the data were binned at 0.015 for the heg and 0.03 for the meg .
we also created background files for the heg and meg spectra using the standard ciao tools . in order to avoid pile - up effects , in this second observation
the source was observed in the continuous clocking ( cc ) mode ( cc33_faint ; time resolution 2.85ms ) .
the source was positioned in the back - illuminated acis - s3 ccd on the nominal target position .
a detailed description on the analysis procedures , such as extraction regions , corrections and filtering applied to the source events and spectra can be found in rea et al .
( 2005b ) .
in order to perform a timing analysis we corrected the events arrival times for the barycenter of the solar system ( with the ciao axbary tool ) using the provided ephemeris .
we used for the timing analysis only the events in the 0.38kev energy range and the standard _
xronos _ tools ( version 5.19 ) .
one fundamental peak plus one harmonic were present in the power - spectrum .
a period of @xmath56s was detected referred to mjd 53189 .
the pulse profile did not change with respect to the previous detection and the 0.38kev pf is @xmath57% .
being the cc mode not yet spectrally calibrated , the timed exposure ( te ) mode response matrices ( rmf ) and ancillary files ( arf ) are generally used for the spectral analysis .
we defer to rea et al .
( 2005b ) for a detailed description on the extraction procedures of the spectral matrices . was observed with the _ swift _ satellite a few times between 2005 january 29th and march 29th , being a calibration source for the timing accuracy and for the wings of the point spread function of the x ray telescope ( xrt ) .
here we focus on data taken in window timing ( wt ) and photon counting ( pc ) mode longer than 1 ks .
we extracted data from two wt observations .
the extraction region is computed automatically by the analysis software and is a box 40 pixels along the wt strip , centred on source , encompassing @xmath58 of the point spread function in this observing mode .
we extracted photons from pc data from an annular region ( 3 pixels inner radius , 30 pixel outer radius ) in order to avoid pile - up contamination .
we consider standard grades 02 in wt and 012 in pc modes .
background spectra were taken from close - by regions free of sources .
the photon arrival times were corrected to the solar system barycenter .
a period search led to a clear detection of the neutron star spin period at @xmath59s , derived with a phase fitting techniques .
this period is consistent with the extrapolation from known ephemerides at a constant period derivative ( kaspi & gavriil 2003 ; dallosso et al .
we found a pf of @xmath60 , @xmath61 , @xmath62 and @xmath63 in the 0.210 kev , 0.22 kev , 24 kev and 410 kev energy bands , respectively .
detailed results are reported in campana et al .
( 2006 ) . a deep observation of the field , was taken on 2003 june 20th from the very large telescope using the naos
conica adaptive optics .
we defer to israel et al .
2004 for details in the data reduction . in fig.[naco_rxs ]
we present the @xmath64 band field around the chandra @xmath49 position of ( israel et al .
2004 ) . besides sources a and b proposed by israel et al .
( 2003b ) , safi harb & west ( 2005 ) and durant & van kerkwijk ( 2006 ) , as the possible ir counterparts to this axp , many further faint sources ( e.g. star c and star d ) were detected strongly consistent with the chandra uncertainty region . unfortunately , @xmath65 band images were obtained under poor sky conditions , and thus the @xmath66 colour could not be determined for these other faint objects .
nonetheless , we note that the @xmath67 magnitudes of e.g. stars c and d ( @xmath68 and @xmath69 , respectively ) are in better agreement with the ir magnitudes typical of axps , than star a and b previously proposed ( @xmath70 and @xmath71 , respectively ) .
deep images in the @xmath72 band were also obtained , but no object was detected within the chandra uncertainty region at a limiting magnitude of @xmath73 , the deepest limit ever obtained for an axps in this band .
detailed results and discussion will be reported in israel et al .
( 2006 , in prep . )
during the post glitches _ beppo_sax observation in 2001 , evidence for an absorption feature was found ( see fig.[line ] ) .
this feature was not detected during the xmm observation 3 years later , the only observation by now with a comparable statistics .
newton upper limit for the line depth is 0.15 at 95% confidence level , which compared with the value found by rea et al .
( 2003 ; [email protected] at 90% confidence level ) , leave only a very small chance that the two measurements are consistent .
[ montecarlo ] we then undertook a careful re - analysis of the _
beppo_sax data .
this re - analysis resulted in the finding that the phases at which the absorption line was strongest were given incorrectly in the published version of rea et al .
( 2003 ) , as we noticed earlier : in particular , the line is strongest close to the pulse minimum in the 0.1 - 2kev band ( or the pulse maximum in the 6 - 10kev band ) .
we nevertheless found that the reported estimate of the significance is sound and not much influenced by different choices in the background subtraction ( annular regions , circular regions far from the source or using blank field files ) or by different extraction regions for the source .
the re - analysis of the _
beppo_sax data made varying the extraction radius , the criterion for the background subtraction and the spectral binning factor , results in basically the same line properties , which strengthen our confidence in the robustness of the result . using an f - test method and taking into account the six trials we made in the phase resolved spectra
, we derive a confidence level for the absorption line of @xmath75 . note that even if we take into account all the possible energies at which the feature could lie in the lecs plus mecs energy range , the confidence level is still @xmath76 .
however , despite being the most common method in astrophysics to derive the significance of the emission or absorption spectral features , protassov et al .
( 2002 ) pointed out that the f - test may be inappropriate in these circumstances , leading sometimes to incorrect significance estimates . following the recipe of protassov et al .
( 2002 ) , in order to further investigate on this significance issue , we ran a monte carlo simulation of @xmath77 spectra fixing only the continuum model ( parameters reported in rea et al .
2003 ) and the same number of photons of the phase resolved spectrum which showed the line in the 2001 _ beppo_sax observation .
the results of the simulation is shown in fig.[montecarlo ] : in the left panel each red circle represents one of the @xmath78 simulated spectra for which a line was detected , here we plot the depth of the lines as a function of the line energy . on the other hand , in the right panel we report the the number of spectra , among the @xmath78 simulated spectra , for which the statistical fluctuation reported the presence of a line at 8.1kev , as the function of the line depth . from this simulation we found 32 spectra with depth @xmath790.8 in @xmath77 points .
we can then reliably say that the probability of the line being a fluctuation is @xmath800.32% . in summary
, we confirm the detection of the 8.1kev absorption line in the _ beppo_sax data made by rea et al .
( 2003 ) at 99.68% confidence level ( see fig.[montecarlo ] ) .
note that the non homogeneous coverage of red circles over the entire spectral range mirrors the energy dependency of the _ beppo_sax spectral matrices .
the interpretation of the absorption feature as a cyclotron scattering line proposed by rea et al .
( 2003 ) was based on the following criteria : 1 ) a gaussian line gives a bad fit and does not reproduce the asymmetrical shape of the observed feature ; 2 ) the best fitting model is the _ xspec cyclabs _ model ; 3 ) the line strength is highly phase dependent ; 4 ) no atomic edges or absorption lines are known to lie around 8.1kev ( at least without assuming ad hoc shifts possibly due to the high gravitational redshift or to zeeman effects in such strong magnetic field ) ; 5 ) the relation between the line energy and width agrees with that of cyclotron scattering features discovered in other classes of sources ( see fig.5 in rea et al .
2003 ) ; 6 ) the magnetic field inferred from the line energy , either in the case of an electron or proton cyclotron resonance , is reasonably consistent with what is expected for a normal neutron star ( @xmath81 g ) or for a magnetar ( @xmath82 g ) , both being still open possibilities . then ,
if this feature is real , all the above points hint toward the cyclotron nature of the absorption feature at 8.1kev . keeping always in mind the possibility that the absorption line in the _
beppo_sax spectrum might be due to statistical fluctuations , in rea et al .
( 2005a ) we discuss physical mechanisms which could be responsible for the appearance of a transient cyclotron line in this source in the context of the magnetar scenario .
the long - term evolution of the source flux and the spectral hardness are shown in fig.[plot_all ] , where different observations , spanning nearly ten years , were collected . comparing the two panels
, there seems to be a correlation between the photon index the source x - ray flux .
the spectrum became progressively harder as the flux rose in correspondence of the two glitches and then softened as the luminosity dropped , following the glitch recovery .
this is suggestive of a scenario in which the mechanism responsible for the glitches is also at the basis of the enhanced emission and of the spectral hardening .
the similarity of the second glitch of with the one discovered during the bursting activity of ( kaspi et al .
2003 ) , after which a similar exponential recovery was seen , suggests that bursts likely occurred in as well , but the sparse observations did miss them , as already suggested by kaspi & gavriil ( 2003 ) .
moreover , the spectral parameters and the flux changes after the recovery of the glitch strengthens this idea in comparison with what was reported for the post - bursts fading of ( woods et al .
2004 ) .
this may be interpreted as the onset of a twist , which grew , culminated in the glitches , and then decayed . a twisted external field , in fact ,
is in an unstable magnetostatic equilibrium and evolves towards a pure dipole field which represents the configuration of minimal energy ( see rea et al .
2005a for a detailed interpretation ) .
the last chandra and swift observations show that the source is slowly increasing again its flux and hardening its emission .
whether the suggested correlation with the glitching and possibly bursting activity holds , we expect the source to re - enter an active state around 20062007 .
if confirmed for all the magnetars , the x - ray monitoring might be an excellent tool to foresee the activity of magnetars .
based on the deep vlt
naco observation , we believe that the identification of the ir counterpart of 1rxs j170849 400910 is still an open issue , mainly due to the very crowded region in which this source is located .
note that the @xmath83 variability recently reported for source b by durant & van kerkwijk ( 2006 ) does not seem to be a conclusive word on the ir counterpart of this source , especially considering that source b has ir magnitudes much brighter than all other axps .
following protassov et al .
( 2002 ) we re - addressed the issue of the significance of the absorption line discovered by _
we end up with a 99.68% confidence level for its existence .
the fact that the source was not completely recovered by the second glitch at the time the line appeared , make the correlation between line appearance , glitching activity and flux enhancement possibly intriguing and might suggests that the conditions for line formation were met at the epoch of the _
beppo_sax pointing .
future long x - ray observations are needed in order to follow the source in its slow flux increase and possibly re - detect the source in such state .
we thank gordon garmire for having observed with _
chandra _ within his guarantee time , and cees bassa for having recognised in the different catalogues used for the astrometry , the reason for the shift between our ir naco field and the one reported by durant & vankerkwijk ( 2006 ) .
we also acknowledge f. haberl , l. kuiper and the anonymous referee for useful comments . | we report here on x ray and ir observations of the anomalous x - ray pulsar ( axp ) .
first , we report on new xmm - newton , swift - xrt and chandra observations of this axp , which confirm the intensity hardness correlation observed in the long term x - ray monitoring of this source .
these new x - ray observations show that the axp flux is rising again , and the spectrum hardening .
if the increase of the source intensity is indeed connected with the glitches and a possible bursting activity , we expect this source to enter in a bursting active phase around 20062007 .
second , we report on deep ir observations of 1rxsj170849400910 , taken with the vlt - naco adaptive optics , showing that there are many weak sources consistent with the axp position .
neither star a or b , as previously proposed by different authors , might yet be conclusively recognised as the ir counterpart of 1rxsj170849400910 .
third , using monte carlo simulations , we re - address the calculation of the significance of the absorption line found in a phase - resolved spectrum of this source , and interpreted as a resonant scattering cyclotron feature . |
radiative transfer , the process that describes radiation propagating through and interacting with matter , is a general problem that is encountered in all areas of astronomy ( and far beyond ) . due to its high dimensionality and nonlocal and nonlinear behaviour ,
it is generally considered as one of the most challenging problems in numerical astrophysics .
recent years have seen an impressive advancement of three - dimensional ( 3d ) radiative transfer studies , thanks to the increase in computational power , the availability of a wealth of new observational constraints and the development of new algorithms . among the different approaches available to solve the radiative transfer problem ,
the monte carlo method is generally the most popular one .
the first monte carlo radiative transfer codes were developed more than four decades ago , and since then the method has steadily increased its market share in all fields where 3d radiative transfer is important .
many powerful monte carlo codes are available in various fields of computational astrophysics , including dust radiative transfer , ly@xmath0 line transfer , ionising radiation transport @xcite , neutron transport @xcite and neutrino radiation transport @xcite
. these and many other papers and monographs discuss at length the main algorithms behind monte carlo radiative transfer and the various proposed improvements and acceleration techniques . a very different aspect of monte carlo radiative transfer codes that has received very little attention in the literature , is the setup of a suite of components that describe the distribution of the sources and sinks in the radiative transfer code ( i.e. , the objects that add radiation to or remove radiation from the radiation field ) .
this is an aspect outside the real core of monte carlo radiative transfer problem .
virtually all of the available codes require the user to hard - code the model makeup for each distinct problem ( although the results of hydrodynamical simulations can generally be processed out of the box , given the appropriate import module ) .
we argue , however , that it is very useful for a radiative transfer code to provide a suite of input models , or _
geometries _ , as built - in components .
such toy models can provide a low - threshold introduction to new users .
they also have a crucial role in a more scientific way : toy input models are used to benchmark different codes , to investigate the effects of dust attenuation on the apparent structural parameters of galaxies , or to investigate the physical impact of e.g. clumpiness or spiral arms .
finally , they are essential for so - called inverse radiative transfer , i.e. when a parameterised radiative transfer model is fitted to observational data . in order to be useful for these goals
, the suite of input models should be diverse enough and contain models with different degrees of complexity , ranging from the metaphorical spherical cows to more realistic toy models that consider , for example , spheroidal and triaxial geometries , including bars , spiral arms and clumpy distributions .
setting up such a suite is more complex than it might appear at first sight .
while each model is in principle completely determined by the 3d density distribution @xmath1 , the setup requires more than just an implementation for this density function .
a crucial aspect of monte carlo radiative transfer codes is the emission of a multitude of simulated photon packets from random locations sampled from the source density distribution .
so each model in the suite should contain a routine that generates random numbers according to @xmath1 .
moreover , this random position generation needs to be very efficient , since the random positions are sampled extremely often . in this aspect ,
monte carlo radiative transfer simulations differ from other codes where random positions need to be sampled from arbitrary density distributions , e.g. to generate the initial conditions for n - body or hydrodynamical simulations .
the efficient generation of random positions from complex 3d density distribution is not a trivial task .
another aspect that needs consideration is the organisation of such a suite of input models .
one could provide a parameterised model that can cover every possible option , but this would quickly lead to an explosion of options that is hard to overview and maintain , and would inevitably contain substantial code duplication . in this paper , we describe how these issues are dealt with in the publicly available 3d monte carlo dust radiative transfer code skirt .
while we summarise the relevant information for the discussion here , an in - depth description of the architecture and overall design of the skirt code can be found in .
it should also be noted that the presented design issues , while discussed in the context of skirt and thus dust radiative transfer , are fully applicable to other monte carlo transport problems .
this paper is laid out as follows . in section [ techniques.sec ]
we review the general techniques to generate random vectors from multivariate distributions that are available in the specialised numerical analysis literature , but less so in the astrophysics community . in section [ geometrysetup.sec ] we present the general setup of the suite of models for the skirt code . in section [ buildingblocks.sec ]
we describe the various building blocks currently present in the skirt code , and in section [ decorators.sec ] we present a number of decorators that can be used to combine and add complexity to simple building blocks .
some decorators are analysed in more detail , focusing on the methods used to efficiently generate random positions . in section [ discussion.sec ]
we discuss the advantages of the decorator - style design of our suite of components , and we critically investigate an alternative option in which random positions are generated using a generic routine rather than customised generators .
we show that , while such an approach is simpler to implement and might hence seem an attractive alternative , it can not compete with our approach , in terms of accuracy and efficiency .
section [ conclusions.sec ] sums up .
the generation of random numbers from univariate distributions is a well - known topic in numerical analysis .
a number of standard methods are widely used and clearly described in standard textbooks ( e.g. , * ? ? ?
additional techniques that are less known in the ( astro)physics community include the acceptance - complement method @xcite and the forsythe - von neumann method @xcite .
for an extensive overview of random number generation from univariate distribution , see @xcite .
unfortunately , the generation of random vectors from multivariate distributions is much more complex .
the only multivariate distributions from which random vectors can be generated directly are those where the density can be written as a product of independent univariate density distributions . in general , however , more advanced techniques are necessary .
the inversion method , also known as inversion sampling , is the most popular method for univariate generation problems .
the basis of this method is the following : if @xmath2 is distributed according to a density @xmath3 , then the variable @xmath4 defined as the solution of the equation @xmath5 is distributed according to the density @xmath6 if we now want to generate a number from a given density @xmath7 , we can take a uniform distribution @xmath8 and set @xmath9 where @xmath10 is the cumulative distribution corresponding to the density @xmath7 . the inversion method can hence be used to generate random variates with an arbitrary density , provided that the inverse function @xmath11 to the cumulative distribution @xmath12 is explicitly known .
classical examples include the exponential distribution , the cauchy distribution , the rayleigh distribution and the logistic distribution .
formula ( [ transfmethod1d ] ) can directly be expanded to multiple dimensions : if @xmath13 is distributed according to the joint probability distribution @xmath14 , then the vector @xmath15 defined as the solution of the vector equation @xmath16 is distributed according to the joint distribution @xmath17 where @xmath18 is the jacobian determinant .
probably the most famous application of this formula is the box - muller method to generate random normally distributed deviates @xcite .
another interesting special application is the case of a linear transformation . in this case
, the transformation @xmath19 can be written as a matrix multiplication @xmath20 , and the distribution of @xmath15 is @xmath21 with @xmath22 the absolute value of the determinant of @xmath23 .
this kind of transformation is particularly useful for the generation of random vectors with a given dependence structure , as measured by the covariance matrix @xcite . in general , however , it is not straightforward to use this identity ( [ transfmethodmd ] ) to construct a method that can be used to generate random positions from an arbitrary multidimensional distribution .
apart from the inversion method , the rejection method , also known as the acceptance - rejection method , is the most popular method to sample nonuniform random numbers from univariate distributions @xcite .
the basic idea behind the method is that , if one wants to sample a random number from a function @xmath7 , one can sample uniformly from the 2d region under the graph @xmath7 .
more concretely , assume that @xmath24 , where @xmath25 is another distribution from which random numbers are easily generated , and @xmath26 is the so - called rejection constant ( it obviously satisfies the condition @xmath27 ) .
one then generates a uniform deviate @xmath28 and a random number @xmath4 from the reference distribution @xmath25 , and calculates the quantity @xmath29 .
this procedure is repeated until @xmath30 , in which case @xmath4 is the desired random number .
one of the advantages of the rejection method is that it does not require that the cumulative distribution function be analytically known , let alone be invertible . however , its effectiveness depends on how accurate @xmath31 is approximated from above by @xmath32 .
less accurate approximation leads to a greater chance of rejection ; on average @xmath26 iterations of the loop are required before one successful random number is generated . moreover
, the reference distribution should be such that random numbers can be easily generated from it and that the computation of @xmath33 is simple . for a range of standard distributions , such as the gamma distribution and the poisson distribution ,
efficient reference functions can be constructed .
the rejection method is by no means limited to univariate distributions , and can immediately be applied to multivariate generation problems .
however , the rejection rate typically increases rapidly when going from one to more dimensions , which decreases the efficiency of the method .
moreover , the design of a suitable reference distribution becomes much more complicated .
the composition method or probability mixing method is another important technique in both univariate and multivariate random number generation @xcite . rather than a method on itself to generate non - uniform random numbers , it is a principle to facilitate and speed up other random number generating methods . the simple idea behind composition is to decompose the distribution @xmath34 as a weighted sum , @xmath35 with all @xmath36 normalised densities , and the weights @xmath37 a probability vector ( i.e. , all @xmath38 and @xmath39 ) . to generate a random @xmath15 from the distribution @xmath34
, we first generate a random integer number @xmath40 from the discrete probability vector @xmath37 , and subsequently generate a random @xmath15 from the density @xmath41 .
a prerequisite for this method to work is that , obviously , the decomposition can be done efficiently , and that the complexity of the problem is reduced by the decomposition .
finally , a powerful technique that applies only to multivariate random number generation is the so - called conditional distribution method @xcite .
it is based on the bayesian identity @xmath42 which expresses that a joint distribution of two variables can be calculated as the product of the marginal distribution of the former variable and the conditional distribution of the latter .
expanding it to multiple dimensions , @xmath43 the beauty of this technique is that it reduces the multidimensional generation problem to a sequence of independent univariate generation problems .
the main drawback is that it can only be used when much detailed information is known about the distribution @xmath34 . in particular , it is by far not always the case that marginal and conditional distributions are easily calculated in closed form .
the standard textbook example of this technique is the multivariate cauchy distribution .
skirt is a multi - purpose 3d monte carlo dust radiative transfer code that is mainly used to simulate dusty galaxies and active galactic nuclei .
the code was designed as a highly modular software , with a particular consideration for the development of a flexible and easy user interface and the use of proven software design principles as described by .
the skirt code offers a wealth of configurable features that are ready to use without any programming .
in particular , the skirt code is equipped with a suite of input model components , the so - called classes , that can be used to represent distributions of either sources or sinks .
essentially , the suite consists of a number of classes that all inherit from an abstract class , for which the c++ class interface looks like .... class geometry { public : geometry ( ) ; virtual ~geometry ( ) ; virtual double density(position x ) const = 0 ; virtual position generateposition ( ) const = 0 ; } .... both the and functions are pure virtual functions , which means that each model in the suite should provide a routine that implements the ( normalised ) density @xmath1 , and a routine that generates random positions according to this density . for example , the interface of a concrete class that contains only a single parameter @xmath44 would look as follows .... class concretegeometry : public geometry { public : concretegeometry(double p ) { _ p = p } double density(position x ) const ; position generateposition ( ) const ; private : double _ p ; } .... for the design of this suite of input models , a naive option would be to provide a set of parameterised models that contain free parameters to control every possible option .
a typical standard component of such a suite would be a very generalised plummer model , with free parameters setting the location of the centre , the orientation with respect to the coordinate system , the scales , the flattening describing potential triaxiality , possible degrees of clumpiness , etc .
this approach has a number of strong disadvantages .
clearly , it would quickly lead to an explosion of different options that is hard to overview .
it would inevitably lead to substantial code duplication ( nearly all the code would need to copied if we would consider a hernquist profile instead of a plummer profile ) and code that is virtually impossible to maintain ( the code for all models would need to be updated if a new feature is added or altered ) . also concerning efficiency
, such a design would not be optimal .
indeed , a random position generating routine for such a generalised plummer model is hard to construct and is certainly much less efficient than the simple routine that is possible for a plain spherical plummer model ( using the inversion method ) . to overcome these problems
, we adopted a completely different approach that is much simpler but still provides the flexibility and functionality needed to set up complex models .
this is achieved using a combination of simple base models on the one hand , and so - called decorator geometries on the other hand .
decorator geometries are not real models on their own , but they apply modifications upon other models in interesting ways , following the _ decorator _ design pattern . in object - oriented programming ,
a decorator attaches additional responsibilities to an object dynamically and provides a flexible and powerful alternative to subclassing for extending functionality @xcite . in our present context
, a decorator geometry is a special kind of geometrical model ( i.e. , it is also a c++ class in the general suite ) that takes one or more other components and adds a layer of complexity upon them .
simple examples of decorators that can easily be implemented are the relocation of the centre of a given component , or the rotation of a given model with respect to the coordinate system .
more complex decorators deform a spherical model to a triaxial one , or add a spiral perturbation to an axially symmetric model .
the advantage of the decorator approach is clear : each decorator needs to be implemented only once , and can then be applied to any possible model . in the following two sections we describe the building blocks in the skirt suite , and a number of decorator geometries that can alter these building blocks to more complex structures .
the skirt suite contains a limited number of elementary input models , which can be used either as elementary toy models , or as building blocks for more complex geometries . for each of them
, the density can be expressed as a simple analytical function , and the generation of random positions reduces to three independent univariate generation problems .
apart from these analytical components , skirt also offers the possibility to set up components in which the geometry of sources and/or sinks is defined by means of particles or on a grid .
in particular , skirt can import a snapshot from a ( magneto)hydrodynamical simulation .
one obvious goal would be to post - process a hydrodynamical simulation in order to calculate the observable multi - wavelength properties of the simulated objects ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in this case
, it would be sufficient to just read in the geometry of the snapshot as it is , and start the radiative transfer simulation .
more generally , however , it would be useful if these numerical components would be at a similar level as the analytical components .
this would open up the possibility to decorate them and combine them with other analytical and/or numerical components to more complex models .
a first group consists of spherically symmetric models . in spherical symmetry
, the generation of a random position from the 3d density @xmath45 simplifies to the generation of a random azimuth from a uniform distribution , a random polar angle from a sine distribution , and a random radius from the univariate density @xmath46 .
the skirt suite includes the most popular models used to represent shells , star clusters , early - type galaxies , or galaxy bulges , such as a power - law shell model @xcite , the plummer model @xcite , the @xmath47-model @xcite , the srsic model , and the einasto model . within this family of models , many other famous profiles
are contained , including the hernquist model @xcite , the jaffe model @xcite and the @xmath48-model @xcite .
a second group of elementary input models consists of axisymmetric models in which the density is a separable function .
the standard example with a density distribution separable in cylindrical coordinates is the double - exponential disc that is the de facto standard to represent disk galaxies in radiative transfer studies .
how random positions can be drawn from this distribution is discussed in appendix a of @xcite .
an example of an axisymmetric model where the density is separable in spherical coordinates is the torus model that is often used to represent the dusty tori around stars or active galactic nuclei @xcite .
the first group of numerical components in skirt are defined as a set of smoothed particles .
this is mainly useful when we want to use the output of a smoothed particle hydrodynamics simulation . in spite of claims that the technique suffers from fundamental problems
@xcite , it is still the most popular hydrodynamics technique , especially for cosmological simulations of galaxy formation ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the output of an sph snapshot consists of a set of `` particles '' ( or rather anchor points in a co - moving grid ) , each characterised by a large suite of physical quantities .
as far as skirt is concerned , most of these physical quantities are irrelevant .
an component in skirt is essentially defined by a list of @xmath49 smoothed particles and the assumed smoothing kernel @xmath50 , with each smoothed particle characterised by a position @xmath51 , a fractional mass @xmath52 and a smoothing length @xmath53 . the total density at an arbitrary position @xmath15 is then given by @xmath54 in practice , the kernels used in sph simulations almost always have a finite support , and in this case only a relatively small number of terms in the sum have a non - zero contribution .
the class in skirt employs smoothing kernel implementations optimised for each specific task .
the geometry s density at a given location is calculated according to equation ( [ rho - sph ] ) using a finite - support cubic spline kernel , so that the operation can be limited to particles that potentially overlap the location of interest . to facilitate this process
, the setup phase of the simulation places a rough grid over the spatial domain and constructs a list of overlapping particles for each grid cell . as described in section 3.3 of ,
a further optimisation is provided to calculate the mass within a given box ( a cuboid lined up with the coordinate axes ) , as an alternative to sampling the density in various locations across the volume of the box . in this case
, the calculation uses the analytical properties of a scaled and clipped gaussian kernel , designed to approximate the cubic spline kernel , to directly determine the mass in the box .
this optimisation accelerates the density calculation for typical cartesian grids by an order of magnitude . on the other hand ,
the generation of random positions sampled from the geometry s density distribution is rather straightforward , thanks to the composition method .
the first step is the choice of a random smoothed particle , based on a discrete distribution where each particle is weighted by its relative mass contribution .
the second step is generating a random position according to the distribution of the chosen particle . the current implementation samples a random number from a gaussian smoothing kernel with infinite support using the inversion method . in the future
we may provide a suite of smoothing kernels , allowing the user to select the kernel that best fits the sph snapshot being imported .
the methods used for calculating the density and generating a random radius would obviously depend on the actual shape of the kernel .
in addition to what is described above , skirt has facilities to associate a spectral energy distribution with each sph star particle based on properties such as age and metallicity , or to associate a dust mass with each sph gas particle based on properties such as temperature and metallicity . in the former case
, the geometry will weigh the particles by luminosity , for each wavelength separately , and in the latter case , the geometry will weigh the particles by dust mass .
a detailed description of these features is beyond the scope of this paper .
apart from smoothed particle hydrodynamics , the main other technique that is used to perform hydrodynamics simulations is eulerian mesh - based hydrodynamics @xcite .
a fundamental ingredient of this technique is the use of grids with adaptive mesh refinement .
eulerian amr simulations are used in virtually all fields of astrophysics , and monte carlo codes have been adapted to work directly on these hierarchical grids .
the in skirt reads in snapshots defined by density fields discretised on hierarchical cartesian grids and converts them to the format of the other components in the skirt suite . calculating the density at an arbitrary position comes down to identifying the cell that contains this position and returning the density associated with that cell . as hierarchical grids have the structure of a tree , isolating the correct cell is a straightforward and computationally cheap operation . generating random positions from such
a component is similar to the case of the smoothed particles , and is based on the composition method .
we first generate a random cell from a discrete distribution where each cell is weighted by its relative mass . secondly
, we determine a random position within the chosen cell ; as the cells in a hierarchical cartesian grids are cuboids , this is a trivial task .
recently , a new lagrangian technique that solves the hydrodynamics equations on a moving , unstructured voronoi grid is gaining popularity . it is claimed to avoid some of the difficulties of smoothed particle hydrodynamics on the one hand and eulerian grid - based hydrodynamics on the other hand .
this technique has been used for many years in the computational fluid dynamics community @xcite , and a number of novel codes based on this principle have recently been developed in the astrophysics community @xcite . moving mesh hydrodynamics
is mainly applied to simulations of galaxy formation and evolution ( e.g. , * ? ? ?
* ; * ? ? ?
skirt contains a class that converts a snapshot from a voronoi hydrodynamical simulation ( or any other density field defined on a voronoi grid ) to a regular skirt component . due to the nature of a voronoi grid ,
the only necessary input is the list of all the generating sites and the associated densities ; it is hence not necessary to store all the vertices and sides of each of the cells .
based on the generating sites , skirt constructs the corresponding unique voronoi grid using the public voro++ library @xcite .
the density routine essentially works in the same way as for the case of the hierarchical grids : it comes down to identifying the cell that contains the given position and returning the density associated to this cell . in the case of a voronoi grid
, however , the cell identification is not as simple as in a hierarchical tree structure of cartesian grid cells . due to the nature of voronoi grids , this is essentially a nearest neighbour search . rather than looping over all possible sites , skirt implements an approach using cuboidal blocks , as explained in detail in .
this task could be optimised even further using more advanced techniques based on space partitioning structures such as @xmath40-d trees or r - trees @xcite .
also the generation of random positions works essentially in the same way as for hierarchical grids .
the first step is identical : we generate a random cell from a discrete distribution where each cell is weighted by its relative mass contribution .
the second step , generating a random position from within the chosen cell , is significantly more complex than in the case of a cuboidal cell . to the best of our knowledge ,
there are no dedicated techniques to generate a random point from a voronoi cell .
there are two possible options .
the first option is to partition the cell into a set of tetrahedra , subsequently select a random tetrahedron from a discrete distribution where every tetrahedron is weighted by its relative volume , and finally generate a random position from the selected tetrahedron .
specific algorithms are available for both the tetrahedrisation of convex polyhedra @xcite and the generation of random positions from a tetrahedron @xcite . the second option , which is more simple and which we have adopted in skirt , is to use the rejection technique . as the reference distribution
we use a uniform density in a cuboidal volume , defined as the 3d bounding box of the cell . as voronoi cells
are convex polyhedra , this bounding box is directly obtained when the vortices of the cell are known ( these are calculated using the voronoi grid setup ) .
extensive testing has shown that , depending on the distribution of the generating voronoi sites , the average ratio of the volume of the bounding box of a voronoi cell over the actual cell volume is about 3 to 4 .
this ratio immediately represents the average rejection rate for the random position generation .
.an overview of the geometry decorators currently implemented in skirt , referencing the section in which they are presented .
geometry decorators can be applied to basic geometry building blocks , and can be chained and combined to create complex structures . [ cols= " < , < ,
< " , ] in this section we describe a number of decorator geometries that can be applied on the building blocks described in the previous section in order to convert them to more complex structures ; see table [ decorators.tab ] for an overview .
the implementation of the density of a decorator geometry is usually not a major problem ; the main challenge is to implement the routine that generates random positions from a decorator geometry , so this is what we focus on .
the decorator in skirt applies an arbitrary offset @xmath55 to any density distribution .
if the original density is @xmath56 , the new density is simply @xmath57 .
generating random positions is equally simple : we generate a random @xmath13 from the original density distribution and return @xmath58 .
the c++ implementation of the class in skirt looks like : .... class offsetgeometrydecorator : public geometry { public : offsetgeometrydecorator(geometry * g , position a ) { _ g = g ; _ a = a ; } double density(position x ) const { return _ g->density(x - a ) ; } position generateposition ( ) const { return _ g->generateposition()+a ; } private : geometry * _ g ; position _ a ; } .... similarly , the decorator rotates any density distribution .
if the rotation is characterised by the orthonormal matrix @xmath23 , the new density is @xmath59 . to generate a random position from this new density ,
we generate a random @xmath13 from the original density and rotate it over the inverse rotation matrix , i.e. @xmath60 .
a third simple decorator , the , carves out a cavity from another density @xmath56 . in formula form
, we have @xmath61 with @xmath62 the fraction of the mass of the original density located in the cavity .
this decorator is useful to represent density distributions of dust close to a star or active galactic nucleus , where the dust has been cleared due to sublimation . to generate random positions from this new density distribution
, we just generate a random position from the original density distribution and reject is when it is located in the cavity .
this is , in fact , an almost trivial application of the rejection technique , where the original density assumes the role of the reference function , and the rejection constant is @xmath63 .
the combines two density distributions into a single distribution according to @xmath64 with @xmath65 the original distributions and @xmath66 their respective weights in the composite distribution . generating random positions for this new density distribution is a trivial application of the composition method . as a first more complex decorator
, we consider the case of a triaxial decorator geometry , which converts a spherically symmetric density distribution into one that is stratified on concentric and confocal ellipsoids .
more concretely , assume that we have a spherically symmetric density distribution @xmath67 , we then consider its triaxial counterpart @xmath68 such triaxial models are discussed and used extensively to describe the stellar distribution of elliptical galaxies and galactic nuclei and the shape of galactic haloes ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
oblate and prolate spheroidal distributions are just a special case of these triaxial models in which @xmath69 .
it is clear that the density ( [ tri : density ] ) can not be written as a product of independent univariate density distributions .
we can use the conditional distribution method . in order to develop a general recipe for generating random positions ,
we start by rewriting the probability distribution according to ( [ tri : density ] ) in spherical coordinates , @xmath70 according to formula ( [ cdm ] ) , we now rewrite this expression as @xmath71 after some calculation , one finds for the marginal distribution for @xmath72 the simple expression @xmath73 note that this marginal distribution is independent of the specific shape of the density profile and only depends on the flattening parameter @xmath44 ( and for @xmath69 , it reduces to a simple uniform distribution , as expected for a spheroidal distribution ) . generating a random azimuth from this density can be done using the standard inversion method ; the corresponding cumulative distribution is @xmath74 and is readily inverted .
once this random @xmath72 has been determined , we can generate a random polar angle @xmath75 from the conditional distribution @xmath76 where we have set @xmath77 .
also this distribution is independent of the specific shape of the density profile and only depends on the flattening parameters @xmath44 and @xmath78 ( and the already randomly determined azimuth ) .
again , generating a random polar angle from this distribution can be done using the standard inversion method : the corresponding cumulative distribution is @xmath79 and is again readily inverted .
finally , we consider the conditional distribution for @xmath80 , @xmath81 if we apply a linear transformation from @xmath80 to @xmath82 the corresponding density for @xmath83 is simply @xmath84 which is nothing but the distribution for a random radius from the original density @xmath56 .
the last step in the generation of a random position is to generate a random position @xmath13 from the original spherical density distribution @xmath56 , and transform the radius @xmath85 of this position to a new radius @xmath80 using the linear transformation ( [ mr ] ) , in which we use the previously determined values for @xmath75 and @xmath72 .
while this method based on conditional probabilities has a certain beauty , there is , actually , a simpler and more efficient method that is based on the inversion method . indeed , consider the simple transformation @xmath86 or in matrix format @xmath87 according to the law of linear transformations ( [ transfmethodlin ] ) , if @xmath13 is distributed according to the spherical density @xmath56 , @xmath15 will be distributed exactly according to the triaxial density ( [ tri : density ] ) . an easy way to generate random positions from a triaxial counterpart of a spherical symmetric density is thus simply generating a random position @xmath13 from the spherical density distribution , from which the desired position can easily be calculated as @xmath60 . as a second complex decorator ,
we consider a logarithmic spiral arm perturbation that can be applied to any axisymmetric density distribution , [ spiral ] @xmath88\ ] ] where @xmath89 are the usual cylindrical coordinates , and the function @xmath90 is defined as @xmath91 \label{perturbation}\ ] ] the factor @xmath92 is a normalisation constant that guarantees that the density is normalised , and is given by @xmath93 this general spiral arm perturbation is a generalisation of the models used by and @xcite , and allows for an arbitrary number of arms @xmath94 , pitch angle @xmath44 , spiral perturbation weight @xmath95 , and arm - interarm size ratio ( controlled by the parameter @xmath49 ) .
for @xmath96 the density reduces to a more simple perturbation with equal arm - interarm size ratio , @xmath97 \right\}\ ] ] for the general 3d density distribution ( [ spiral ] ) , we can again use the conditional distribution method , and write @xmath98 since the spiral perturbation is such that it averages out along every single circle around the @xmath99-axis , i.e. , @xmath100 we trivially find @xmath101 this is exactly the 2d probability density distribution that corresponds to the density of the axisymmetric density distribution .
we hence simply generate a random position @xmath15 from our original density distribution and save the radius @xmath102 and the height @xmath99 . in order to generate a random azimuth @xmath72
, we consider the conditional distribution @xmath103 , for which we find directly @xmath104 this univariate density is independent of the shape of the original axisymmetric density profile ; it only depends on the parameters of the spiral perturbation and randomly determined radius @xmath102 . for the general expression ( [ perturbation ] ) , the standard inversion technique is not easily applicable ; the cumulative distribution corresponding to the density ( [ spiral : pdf ] ) can be calculated analytically , but the resulting expression can not be inverted analytically .
one could resort to a numerical inversion , but a better approach is to use the rejection technique .
we can use a simple uniform density as the reference distribution , @xmath105 .
the rejection constant @xmath26 should ideally be chosen as the smallest value for which @xmath106 for all values of @xmath72 .
since the maximum value of the perturbation function @xmath90 is @xmath92 , we find @xmath107 actually , as an alternative to the conditional probability technique , we could also simply have used the rejection technique from the beginning .
the 3d density @xmath1 satisfies the condition @xmath108 , so if we simply use the original density @xmath109 as our reference distribution , we have essentially the same algorithm with the same rejection constant . the difference between this 3d rejection technique and the previous version , where we used the rejection technique only for the conditional distribution for the azimuth ,
is a matter of efficiency . in the former version
only one random position needs to be generated from the original density , and it the rejection is only to determine the azimuth . in the latter version ,
an entirely new position is generated for every rejection , which is less efficient .
as a last example , we consider the class that turns any density distribution @xmath56 into a clumpy analogue , i.e. , a density distribution in which a fraction of the mass is distributed `` smoothly '' as in the original density distribution , and the remaining fraction is locked up into compact clumps . the density of this distribution can be written as @xmath110 where @xmath111 is the fraction of the mass in clumps , @xmath49 is the number of clumps , and the positions @xmath112 are the central positions of the @xmath49 clumps , determined randomly according to the original density function @xmath113 .
the function @xmath50 is the smoothing function that sets the distribution of matter within every single clump , with @xmath114 a characteristic length scale .
this can in principle be any spherical 3d distribution ; in practice we use either a uniform density sphere or one of the compact support kernels that are used in smoothed particle hydrodynamics simulations .
it is straightforward to generate random positions from this distribution .
it is a direct application of the composition method , and the case is very similar to smoothed particle building blocks discussed in section [ sph.sec ] .
as indicated in section [ geometrysetup.sec ] , a major advantage of the decorator - style approach is that each decorator needs to be implemented only once , and can then be applied to different models .
this avoids the need to implement complex geometries with many different possible features and heavy code duplication . that alone is already an argument strong enough to justify its use .
another strong advantage is the flexibility offered by this approach .
the decorators discussed in section [ decorators.sec ] transform an input model into a more complex model .
such a decorated model is often still relatively simple : it typically adds one layer of complexity onto the model that is being decorated ( think of a rotated exponential disc or a triaxial srsic model ) . if we want to generate toy models that can be used as idealised representations of , for example , a spiral galaxy , various layers of complexity have to be added .
the decorator - style approach is ideal for this purpose : decorators can be applied also to models that have already been decorated . in other words , decorators can easily be chained or nested .
the left - hand panel of figure [ chaineddecorators.fig ] shows a projected image of a relatively complex toy model , constructed using the suite in skirt .
the model consists of a bulge component and a clumpy spiral disc .
each of these two components have a very simple spherical @xmath47-model as their starting point , on which a chain of decorators was applied .
the former was first decorated into a triaxial model , the latter was turned into an axisymmetric disc by applying a very strong flattening , then a spiral perturbation was added , and a fraction of its mass was turned into clumps . in decoration chains the successive decorators do not necessarily need to be different .
it is also possible to chain the same decorator several times , i.e. to apply them recursively .
a powerful example of this nesting is a repeated application of the .
the right - hand panel of figure [ chaineddecorators.fig ] shows an application of this principle on a simple triaxial plummer model . in this case , we have applied four nested applications of the decorator , in which the number of clumps increases in each level , whereas the smoothing length decreases .
this algorithm results in a fractal density distribution that is self - similar over an order of magnitude in scale @xcite .
the possibility to chain decorators , including a repeated application of the same decorator , facilitates the construction of very complex models out of simple building blocks .
analytical components can easily be combined with numerical components based on smoothed particles or grids , and decorators can be applied regardless of the underlying component type .
it is , for example , possible to add a smooth triaxial bulge to an irregular disc structure defined using sph particles , or to carve out a cavity from a complex hydrodynamics system and locate a small nuclear structure there to simulate the effects of an agn on the large - scale structure of a galaxy .
one could argue that the setup that we have chosen is unnecessarily complex .
an alternative approach would be a design where we only provide the density for each component in the suite , and where the generation of random positions is executed by a generic routine rather than a geometry - specific routine for each component .
such a suite could still use the decorator design pattern , but would be simpler to implement .
the main challenge is the design of a generic routine that can generate random positions corresponding to an arbitrary 3d density function .
this is in principle possible : there are a few multivariate black - box random number generation libraries available , including foam @xcite and ranlip @xcite , based on the so - called composition - rejection method , a combination of the composition and rejection methods . in a first phase , the exploration phase , the domain of a distribution is partitioned into different cells . in the generation phase , the rejection technique is used on each cell , where typically a constant density function is used as the reference distribution .
the advantages of this approach are that , in principle , it can be applied for all distributions and in any dimension , and that only the density of each component is required .
the main disadvantage is the complexity and the overhead , both in memory and run - time , that are linked to the exploration phase . in order to test
whether or not our design choice of providing a customised random position generator for each concrete class ( and for each decorator , in particular ) is justified , we have set up a comparison between our routines and a generic black - box routine .
we have implemented a new decorator class , , that replaces the model - specific random position generator by a generic routine based on the foam library .
foam is a self - adapting cellular code that uses iterative binary splitting , using either simplexes or hyper - rectangles , to subdivide the configuration space .
we set up a simple suite of test models , for which we generate random positions using both the model - specific generator and the generic foam generator .
our suite consists of a sequence of models with an increasing level of complexity .
it essentially follows the different steps in building the model shown in the left panel of figure [ chaineddecorators.fig ] .
the following models are considered 1 . a spherical @xmath115 model , 2 . model # 1 flattened to an axisymmetric disc , 3
. model # 2 with spiral perturbation , 4 . model # 3 with a fraction turned into clumps , 5 .
model # 4 with a triaxial @xmath115 bulge added . for each model
, we generate @xmath116 random positions ( we do not use any photon propagation or detection as would be used in monte carlo radiative transfer simulation in order to isolate the random position generation process ) .
the timings were done using a single core and with averages over multiple test runs .
one crucial aspect for the foam - based generator is the choice of the foam parameters , in particular of the number of cells in the grid
. a larger number of cells implies a higher computational cost of the exploration phase on the one hand , but a better approximation of the density and smaller rejection rates .
the ideal choice of this parameter is impossible to determine in a general way .
we consider 5 different values for the number of cells in the foam for every model , ranging from 25,000 to 400,000 .
the results are illustrated in fig .
[ timings.fig ] , where , for each model , we give the total run time and the contributions for the setup of the foam and the actual time spent on generating the random positions . a first major conclusion that can be drawn from these results is the efficiency of the customised random position generation routines . from a simple spherical model , to a complex model composed of two components that both have been decorated multiple times , the computational cost increases only by a factor two . moving from a spherical to a triaxial model does not affect the efficiency of the generation of random positions at all ( not a surprise , as it just implies two simple multiplications , as shown in section [ triaxial.sec ] ) .
the biggest jump in timing occurs when spiral structure is added to the model ( i.e. , from model # 2 to # 3 ) , because the random position generation is based on the rejection method , which has more overhead compared to the inversion and composition methods .
the addition of clumping ( from model # 3 to # 4 ) hardly implies an increase in computation time .
note that the addition of a bulge even decreases the computation time . in model # 5 half of the positions
are generated according to the density of the relatively simple triaxial bulge , whereas in model # 4 they all are generated according to the more complex clumpy spiral structure .
a second major conclusion is that the generic foam - based random position generator is significantly less efficient than the customised generator . for the most simple models ( # 1 and # 2 ) ,
the difference in speed is a factor 38 , depending on the number of cells used .
the increase in total run time for the foam - based models with increasing number of cells is mainly due to the time necessary to set up the foam .
the time necessary to generate the random positions also increases , but not as strongly . for the more complex models ,
the efficiency of the foam - based generator decreases even more .
the biggest jump in timing now occurs when we add clumping to the model , because the calculation of the density for the clumpy models is computationally very demanding ( each density evaluation requires a sum over a large number of clumps ) . for models # 4 and 5 , the foam - based generators are at least an order of magnitude slower than the customised routines .
interestingly , the run time of these simulations does not increase with the number of cells . for simulations with few cells ,
the setup phase is relatively quick , but the generation phase is very inefficient .
the reason is that the cells are relatively large with a strong variation in density values , and hence the rejection rates are uncomfortably large .
if the number of cells in the grid increases , the setup time naturally increases , but the actual position generation becomes more efficient because the average rejection rates are smaller .
moreover , the inefficiency of the foam - based generator is not the only problem , but also accuracy is an issue , in particular for complex models with multiple local maxima and strong gradients , such as models # 4 and # 5 .
if the number of cells is limited and the cells hence rather large , both the total mass and the local maximum within a cell are very hard to guess ( they are determined by random sampling the cell ) . as a result ,
both the relative weights of the different cells and the rejection constants within the individual cells are poorly determined .
this will lead to an artificial smoothing and wrong results .
figure [ genericversusfoam.fig ] shows a sequence of images corresponding to model # 4 , created using the customised random position generator ( top left ) and using the foam - based one with different values for the number of cells ( remaining panels ) .
the customised generator generates an image that reveals all the details that are expected , including the sharp density gradients and the contrast between arm and interarm regions .
the only limitation in the image is the finite resolution due to the pixel size and the unavoidable poisson noise .
the foam images on the other hand show a clear signature of degradation that gradually decreases when the number of cells grows .
for the grids with 25,000 and 50,000 cells , and even for the one with 100,000 cells , the effects of the grid are clearly visible , and the individual clumps and the spiral structure are insufficiently resolved . for the grid with 400,000 cells , the individual clumps are well resolved , but the sharpness of the spiral arms is still under - resolved , especially in the central regions . since both the accuracy and the efficiency of the generic foam - based random position generator can not compete with the customised versions , we conclude that it is worth investing in the latter , and that our design choice is justified .
we have described the design of a suite of components that can be used to model the distribution of sources and sinks in the monte carlo radiative transfer code skirt .
our main conclusions are the following : * the availability of a well - designed suite of input models , with enough variety and different degrees of complexity , in a publicly available monte carlo code has a strong added value .
such models can serve as toy models to test new physical ingredients , or as parameterised models for inverse radiative transfer fitting .
they also provide a low - threshold introduction to new users , since models with differing degrees of complexity can be run without any coding at all . *
each model is in principle completely determined by the 3d density distribution @xmath1 . in order to be used in monte carlo radiative transfer , however , each model should contain a routine that generates random numbers according to this @xmath1 .
finding the most suitable algorithm to implement this is the most challenging aspect of the design of a suite of input models . *
the design of the skirt suite is based on a combination of basic building blocks and the extensive use of decorators .
the building blocks can either be simple analytical components , or they can be numerical components defined as a set of smoothed particles or on a hierarchical or unstructured voronoi grid .
the decorators combine and alter these building blocks to more complex structures . *
the different multivariate random number generation techniques that exist in the specialised numerical analysis literature can be used to efficiently implement complex decorators , for example those that add triaxiality , spiral structure or clumpiness to other models .
decorators can be chained without problems and essentially without limitation .
different layers of complexity can hence be added one by one .
the result is that very complex models can easily be constructed out of simple building blocks , without any coding at all .
* from the software design point of view , decorators have many advantages , including code transparency , the avoidance of code duplication , and an increase in code maintainability .
this is a clear example that adhering to proven software design principles pays off , even for small and mid - sized projects . *
finally , we demonstrate that our design using customised random position generators is superior to a simpler suite design based on a generic black - box random position generator . using a suite of test simulations with increasing complexity
we demonstrate that our customised random number generators are more accurate and more efficient .
this work fits in the charm framework ( contemporary physical challenges in heliospheric and astrophysical models ) , a phase vii interuniversity attraction pole ( iap ) program organised by belspo , the belgian federal science policy office .
the authors thank ilse de looze and sbastien viaene for their careful reading of a draft version of this paper , and all skirt users for their feedback , which has led to many improvements and additions to the code . , o. , moore , b. , stadel , j. , potter , d. , miniati , f. , read , j. , mayer , l. , gawryszczak , a. , kravtsov , a. , nordlund , . , pearce , f. , quilis , v. , rudd , d. , springel , v. , stone , j. , tasker , e. , teyssier , r. , wadsley , j. , walder , r. , 2007 .
fundamental differences between sph and grid methods . mon . not .
r. astron .
380 , 963978 . ,
m. , davies , j. i. , dejonghe , h. , sabatini , s. , roberts , s. , evans , r. , linder , s. m. , smith , r. m. , de blok , w. j. g. , 2003 .
radiative transfer in disc galaxies - iii .
the observed kinematics of dusty disc galaxies . mon . not .
r. astron .
343 , 10811094 .
, m. , fritz , j. , gadotti , d. a. , smith , d. j. b. , dunne , l. , da cunha , e. , amblard , a. , auld , r. , bendo , g. j. , bonfield , d. , burgarella , d. , buttiglione , s. , cava , a. , clements , d. , cooray , a. , dariush , a. , de zotti , g. , dye , s. , eales , s. , frayer , d. , gonzalez - nuevo , j. , herranz , d. , ibar , e. , ivison , r. , lagache , g. , leeuw , l. , lopez - caniego , m. , jarvis , m. , maddox , s. , negrello , m. , michaowski , m. , pascale , e. , pohlen , m. , rigby , e. , rodighiero , g. , samui , s. , serjeant , s. , temi , p. , thompson , m. , van der werf , p. , verma , a. , vlahakis , c. , 2010 .
herschel - atlas : the dust energy balance in the edge - on spiral galaxy ugc 4754 .
astroph .
518 , l39 . ,
m. , verstappen , j. , de looze , i. , fritz , j. , saftly , w. , vidal prez , e. , stalevski , m. , valcke , s. , 2011 .
efficient three - dimensional nlte dust radiative transfer with skirt .
j. suppl .
196 , 22 . ,
g. , baes , m. , camps , p. , fritz , j. , de looze , i. , hughes , t. m. , viaene , s. , gentile , g. , 2014 . the distribution of interstellar dust in califa edge - on galaxies via oligochromatic radiative transfer fitting . mon . not
r. astron .
441 , 869885 . , i. , fritz , j. , baes , m. , bendo , g. j. , cortese , l. , boquien , m. , boselli , a. , camps , p. , cooray , a. , cormier , d. , davies , j. i. , de geyter , g. , hughes , t. m. , jones , a. p. , karczewski , o. . ,
lebouteiller , v. , lu , n. , madden , s. c. , rmy - ruyer , a. , spinoglio , l. , smith , m. w. l. , viaene , s. , wilson , c. d. , 2014 .
high - resolution , 3d radiative transfer modeling .
i. the grand - design spiral galaxy m51 .
astroph .
571 , a69 .
desbrun , m. , gascuel , m .-
smoothed particles : a new paradigm for animating highly deformable bodies . in : proceedings of the eurographics workshop on computer animation and simulation 96 .
. 6176 . , b. , olson , k. , ricker , p. , timmes , f. x. , zingale , m. , lamb , d. q. , macneice , p. , rosner , r. , truran , j. w. , tufo , h. , 2000 .
flash : an adaptive mesh hydrodynamics code for modeling astrophysical thermonuclear flashes .
j. suppl .
131 , 273334 . , g. l. , danese , l. , 1994 .
thick tori around active galactic nuclei - a comparison of model predictions with observations of the infrared continuum and silicate features .
mon . not .
r. astron . soc . 268 , 235252 . ,
c. , popescu , c. c. , tuffs , r. j. , 2006 . modelling the spectral energy distribution of galaxies .
iv . correcting apparent disk scalelengths and central surface brightnesses for the effect of dust at optical and near - infrared wavelengths .
astroph .
456 , 941952 .
mller , m. , charypar , d. , gross , m. , 2003 .
particle - based fluid simulation for interactive applications . in : proceedings of the 2003 acm siggraph / eurographics symposium on computer animation .
. 154159 . , i. , wolf , s. , steinacker , j. , dullemond , c. p. , henning , t. , niccolini , g. , woitke , p. , lopez , b. , 2004 .
the 2d continuum radiative transfer problem .
benchmark results for disk configurations .
astroph .
417 , 793805 . ,
c. , harries , t. j. , min , m. , watson , a. m. , dullemond , c. p. , woitke , p. , mnard , f. , durn - rojas , m. c. , 2009 .
benchmark problems for continuum radiative transfer .
high optical depths , anisotropic scattering , and polarisation .
astroph .
498 , 967980 . ,
m. , wada , k. , prieto , m. a. , burkert , a. , tristram , k. r. w. , 2014 .
time - resolved infrared emission from radiation - driven central obscuring structures in active galactic nuclei .
mon . not .
r. astron .
445 , 38783891 . ,
j. , crain , r. a. , bower , r. g. , furlong , m. , schaller , m. , theuns , t. , dalla vecchia , c. , frenk , c. s. , mccarthy , i. g. , helly , j. c. , jenkins , a. , rosas - guevara , y. m. , white , s. d. m. , baes , m. , booth , c. m. , camps , p. , navarro , j. f. , qu , y. , rahmati , a. , sawala , t. , thomas , p. a. , trayford , j. , 2015 .
the eagle project : simulating the evolution and assembly of galaxies and their environments .
mon . not .
r. astron .
soc . 446 , 521554 . , m. , fritz , j. , baes , m. , nakos , t. , popovi , l. v c. , 2012 .
3d radiative transfer modelling of the dusty tori around active galactic nuclei as a clumpy two - phase medium .
mon . not .
r. astron .
420 , 27562772 . , j. m. , norman , m. l. , 1992 .
zeus-2d : a radiation magnetohydrodynamics code for astrophysical flows in two space dimensions .
i - the hydrodynamic algorithms and tests . astrophys . j. suppl .
80 , 753790 . ,
i. , asensio ramos , a. , rubio - martn , j. a. , graham , a. w. , aguerri , j. a. l. , cepa , j. , gutirrez , c. m. , 2002 .
triaxial stellar systems following the r@xmath119 luminosity law : an analytical mass - density expression , gravitational torques and the bulge / disc interplay .
mon . not .
r. astron . soc .
333 , 510516 . ,
r. c. e. , van de ven , g. , verolme , e. k. , cappellari , m. , de zeeuw , p. t. , 2008 .
triaxial orbit based galaxy models with an application to the ( apparent ) decoupled core galaxy ngc 4365 .
mon . not .
r. astron .
385 , 647666 . , m. , genel , s. , springel , v. , torrey , p. , sijacki , d. , xu , d. , snyder , g. , nelson , d. , hernquist , l. , 2014 .
introducing the illustris project : simulating the coevolution of dark and visible matter in the universe .
mon . not .
r. astron .
444 , 15181547 . | the monte carlo method is the most popular technique to perform radiative transfer simulations in a general 3d geometry .
the algorithms behind and acceleration techniques for monte carlo radiative transfer are discussed extensively in the literature , and many different monte carlo codes are publicly available . on the contrary
, the design of a suite of components that can be used for the distribution of sources and sinks in radiative transfer codes has received very little attention .
the availability of such models , with different degrees of complexity , has many benefits .
for example , they can serve as toy models to test new physical ingredients , or as parameterised models for inverse radiative transfer fitting . for 3d monte carlo codes
, this requires algorithms to efficiently generate random positions from 3d density distributions .
we describe the design of a flexible suite of components for the monte carlo radiative transfer code skirt .
the design is based on a combination of basic building blocks ( which can be either analytical toy models or numerical models defined on grids or a set of particles ) and the extensive use of decorators that combine and alter these building blocks to more complex structures . for a number of decorators ,
e.g. those that add spiral structure or clumpiness , we provide a detailed description of the algorithms that can be used to generate random positions .
advantages of this decorator - based design include code transparency , the avoidance of code duplication , and an increase in code maintainability .
moreover , since decorators can be chained without problems , very complex models can easily be constructed out of simple building blocks .
finally , based on a number of test simulations , we demonstrate that our design using customised random position generators is superior to a simpler design based on a generic black - box random position generator
. radiative transfer methods : numerical designing software design patterns |
the neutrino signal from the next core - collapse supernova in our galaxy will give us an unprecedented opportunity to peer into the heart of an exploding star and confront our current paradigm of how these stars explode with observations .
but decoding the message will be no easy task because the neutrino signal will have experienced so many flavor - changing events on the trip from proto - neutron star to our detectors that scramble the information , see for example kneller , mclaughlin & brockman @xcite and lund & kneller @xcite .
the first flavor changing effect the signal experiences is due to neutrino self interactions / collective effects in the region up to @xmath5 above the proto neutron star @xcite followed by the mikheyev , smirnov & wolfenstein ( msw ) effect which is complicated in supernovae by the impression of the shockwave racing through the stellar mantle @xcite .
turbulence in the mantle , seeded during the earlier neutrino heating / standing accretion shock instability phase @xcite , also needs to be included usually by modelling @xcite . finally , there is the possibility of earth matter effects leaving an imprint in the signal though a recent study expects this effect to be minimal @xcite .
what makes decoding the signal even more of a challenge is that the neutrinos we receive at a given instant and with a given energy will not have experienced the same flavor evolutionary history .
the neutrinos arriving at a detector will have been emitted from different locations at the source and both the neutrino collective and the msw+turbulence effects will vary from trajectory to trajectory . starting with duan _
@xcite , the self interaction effects in calculations where the neutrino emission over the source is assumed to be spherically symmetric have been seen to be ` angle dependent ' in the sense that a neutrino following a pure radial trajectory differs from one emitted at an angle relative to the normal . presumably allowing for aspherical source emission would only make the trajectory dependence even stronger .
similarly the msw plus turbulence effects are also trajectory dependent .
if we temporarily cast aside the turbulence and focus on the gross structure of the explosion i.e. the lowest angular multipole moments , an aspherical passage of the shock through the star , by itself , leads to a line - of - sight dependence .
but , one must recall that we will not observe the neutrinos from a supernova at widely different lines of sight , all our detectors are here on earth .
the size of the source is of order the proto - neutron star radius , i.e. @xmath6 while the shock effects show up in the signal when the shock has propagated out to @xmath7 .
as long as the curvature of the shock is over a lengthscale greater than the source size the neutrinos which appear in our detectors will all have seen essentially the same profile .
when one re - inserts the turbulence into the profile , one realizes this approximation may no longer be valid because turbulence extends to much smaller lengthscales even when the shock is far out in the stellar mantle .
the density profile along two , essentially parallel lines of sight to a distant detector separated by @xmath6 will no longer be negligibly dissimilar and one must consider how the dissimilarity of the profiles propagates to the neutrinos .
any correlation will lead to a potential new feature of the neutrino signal .
it has been shown that the transition probabilities for a single neutrino - the set of probabilities that relates the initial state to the state after passing through the supernova - is not unique when turbulence is inserted into a profile : it will depend upon the exact turbulence pattern seen by the neutrino as it travelled through the supernova @xcite . those transition probabilities are drawn from distributions whose properties will depend upon the stage of the explosion , the character of the turbulence , and the neutrino energy and mixing parameters . if the coherence of two neutrinos emitted at the same time and with the same energy but from different locations is small then the final states are uncorrelated and one would expect that the flux at a detector would just be the mean of whatever distribution describes the transition probabilities multiplied by the initial spectra .
but if the coherence is high then all the neutrinos will have the same set of transition probabilities which one might expect to ` scintillate ' together as the turbulence evolves .
of course , this ignores the issue of energy resolution and temporal binning of the signal that becomes necessary because of the limited statistics . the purpose of this paper is to consider the issue of finite source size and the correlation of the neutrino transition probabilities along parallel trajectories through turbulent supernova profiles .
our calculations expand upon the work of kneller & volpe @xcite and kneller & mauney @xcite upon which we rely heavily for the techniques used to calculate the turbulence effects and as context for our results .
we begin by describing the calculations we undertook paying particular attention to the construction of the random fields used to model the turbulence .
the basic approach to determining the effects of turbulence are then demonstrated , followed by the computation of the transition probability correlation as a function of the separation between the emission points .
we finish by summarizing our findings and discuss the implications for the galactic neutrino burst signal .
the neutrino transition probabilities are the set of probabilities of measuring some neutrino state @xmath8 given an initial neutrino state @xmath9 i.e @xmath10
. we shall denote antineutrino transition probabilities by @xmath11 .
if the @xmath12-matrix relating the initial and final wavefunctions is known then these probabilities are just the square amplitudes of the elements of @xmath12 .
the @xmath12-matrix is calculated from the schrodinger equation @xmath13 where @xmath14 is the hamiltonian .
the hamiltonian is the sum of the vacuum contribution @xmath15 and the msw potential @xmath16 which describes the effect of matter .
the vacuum hamiltonian is diagonal in what is known as the ` mass ' basis and in this basis @xmath17 is defined by two mass squared differences @xmath18 and the neutrino energy @xmath19 .
the mass basis is related to the flavor basis by the maki - nakagawa - sakata - pontecorvo @xcite unitary matrix @xmath20 .
the most common parametrization of @xmath20 is in terms of three mixing angles , @xmath21 , @xmath22 and @xmath23 , a cp phase and two majoranna phases .
the msw potential @xmath16 is diagonal in the flavor basis because matter interacts with neutrinos based on their flavor .
the neutral current interaction leads to a contribution to @xmath16 which is common to all flavors .
this may be omitted because it leads only to a global phase which is unobservable .
the charged current potential only affects the electron flavor neutrino and antineutrinos and is given by @xmath24 where @xmath25 is the fermi constant and @xmath26 the electron density . in matter the two contributions to @xmath14 means neither the mass nor the flavor states diagonalize the matrix .
but there is a basis known as the matter basis which does diagonalize @xmath14 i.e. for a given value of the electron density there is a matrix @xmath27 such that @xmath28 where @xmath29 is the diagonal matrix of eigenvalues .
when the msw potential vanishes the matter basis becomes the mass basis up to arbitrary phases .
the matter basis is the most useful for studying the evolution of neutrinos through matter because it removes the trivial adiabatic msw transition and it will be the basis we use to report our results in this paper .
we refer the reader to kneller & mclaughlin @xcite and galais , kneller & volpe @xcite for a more detailed description of the matter basis .
neutrino with mixing angle @xmath30 and mass splitting @xmath31 [ fig : profile ] ] we now turn our attention to the turbulent density profiles through which we shall send our neutrinos . as usual , we shall model the turbulence by multiplying a turbulence free density profile by a gaussian random field .
since the spatial extent of the neutrino emission , of order @xmath32 , is much smaller than the radial location of the turbulence , of order @xmath33 , we shall ignore any curvature of the density profile features and use a plane - parallel model for the supernova .
the z axis of our cartesian co - ordinate system is aligned with the radial direction of the profile .
the profile we adopt is from a one - dimensional hydrodynamical simulation of a supernova taken from kneller , mclaughlin & brockman @xcite .
this profile is shown in figure ( [ fig : profile ] ) and is the same one used in kneller & mauney @xcite .
the figure shows the presence of two shocks : the forward shock at @xmath34 and the reverse shock at @xmath35 . in multi - dimensional simulations of supernova
both these shock fronts are aspherical and fluid flow through the distorted shocks leads to strong turbulence in the region between them .
our selection of this profile also determines the neutrino energy we shall use since we wish the neutrinos to have an h resonance density that does not intersect the shocks
. therefore we pick @xmath36 for the neutrino energy and the reader may observe that the two - flavor resonance density for a @xmath36 , shown in the figure , does not intersect the shocks as required .
the turbulence is inserted by multiplying the profile in the region between the reverse and forward shocks by a factor @xmath37 where @xmath38 is a three - dimensional gaussian random field with zero mean . the random field
is represented by a fourier series , that is @xmath39 in this equation the parameter @xmath40 sets the amplitude of the fluctuations while the two @xmath41 functions are included to suppress fluctuations close to the shocks and prevent discontinuities .
the parameter @xmath42 is a damping scale which we set to @xmath43 .
the random part of @xmath44 appears in the the second half of equation ( [ eq : f3d ] ) because the set of co - efficients @xmath45 and @xmath46 are independent standard gaussian random variates with zero mean .
the @xmath47 are a set of wavenumbers and , finally , the paramaters @xmath48 are k - space volume co - efficients .
the method of fixing the @xmath49 @xmath50 s , @xmath16 s , @xmath51 s and @xmath52 s for a realization of @xmath44 is ` variant c ' of the randomization method described in kramer , kurbanmuradov , & sabelfeld @xcite which we have generalized to three dimensions . the randomization method in general partitions the space of wavenumbers into @xmath53 regions and from each we select a random wavevector using the power - spectrum , @xmath54 , as a probability distribution .
the volume paramaters @xmath48 are the integrals of the power spectrum over each partition if the power spectrum is normalized to unity .
variant c of the randomization method divides the k - space so that the number of partitions per decade is uniform over @xmath55 decades starting from a cutoff scale @xmath4 .
throughtout this paper we shall use a wavenumber cutoff @xmath4 set to twice the distance between the shocks i.e. @xmath56 .
the logarithmic distribution of the modes increases the efficiency of the algorithm in the sense that we can use a ` small ' value of @xmath49 and also the agreement between the exact statistical behavior of the field and that of an ensemble of realizations is uniform over some range of lengthscales i.e. it is scale invariant .
this feature is important for our study because the oscillation wavelength of the neutrinos is constantly changing as the density evolves .
the minimum lengthscale we need to cover has been shown by friedland & gruzinov @xcite and kneller , mclaughlin & patton @xcite to be the reduced oscillation wavelengths for the neutrinos and antineutrinos i.e. @xmath57 and @xmath58 - where @xmath59 and @xmath60 are the differences between the eigenvalues @xmath61 and @xmath62 of the neutrinos and antineutrinos respectively .
kneller & mauney @xcite showed the wavelengths in the turbulence region were of order @xmath63 or greater which is approximately four orders of magnitude smaller than the shock separation .
this means we need to pick @xmath64 to cover the necessary decades in k - space .
the final component of our calculations we have yet to discuss is the power spectrum @xmath54 . in this paper
we shall consider two power spectra and our first choice , due to its simplicity , is a normalized three - dimensional , isotropic inverse power - law spectrum given by @xmath65 for @xmath66 where @xmath67 is the magnitude of the wavevector @xmath68 . throughout this paper
we shall adopt the kolmogorov spectrum where @xmath69 .
the one dimensional power spectrum for the @xmath70 component of the wavevector is @xmath71 which differs from the one dimensional power spectrum used by kneller & mauney @xcite because for @xmath72 the power is suppressed by the factor @xmath73 and the one dimensional spectrum is non - zero for @xmath74 .
the two - point correlation function @xmath75 for this choice of a power spectrum depends only the magnitude of the separation , @xmath76 , and may be calculated analytically to be @xmath77 where @xmath78 is the incomplete gamma function .
there is one last quantity to determine : the number of @xmath49 of elements in the sets of random wavenumbers , coefficients and volumes . to find this quantity we compare the statistical properties of an ensemble of random field realizations with the exact expressions as a function of the ratio @xmath79 for a given @xmath55 .
the statistical property we compute is the second order structure function @xmath80 which is given by @xmath81 where @xmath82 is the separation between two points .
the function @xmath83 is related to the two - point correlation function @xmath84 via @xmath85 . for the isotropic power spectrum both @xmath86 and @xmath52
are only functions of the magnitude of @xmath82 and the correlation function is given above .
as a function of @xmath87 for two randomly orientated points in a 3-d homogeneous and isotropic gaussian random field to the exact structure function . the two curves in the figure correspond to @xmath88 ( blue solid ) and @xmath89 .
at every @xmath87 we generated @xmath90 realization of the field and the error bar on each point is the standard deviation of the mean @xmath91 .. [ fig : grf3diso ] ] in figure ( [ fig : grf3diso ] ) we show the ratio @xmath92 of the numerically calculated structure function for the isotropic random field to the exact solution as a function of the scale @xmath93 when we use either @xmath94 wavenumbers spread over @xmath95 decades or @xmath96 wavenumbers over @xmath97 decades .
the numerical calculation is the average of @xmath90 realizations of the turbulence and the error bar on each point is the error on the sample mean .
the figure indicates that the method we use to generate random field realizations reproduces the analytic results for the structure function very well and with high efficiency because good agreement between the statistics of the ensemble and the exact result requires just @xmath98 .
in fact we find even @xmath99 ratios of just @xmath100 are sufficient to give acceptable agreement but we re - assure the reader we shall stick with @xmath98 .
but isotropic and homogeneous three - dimensional turbulence is perhaps not a realistic scenario for supernova because the gravitational potential and the general fluid flow are in the radial direction . only on sufficiently small scales
should the turbulence become isotropic .
this division into large and small lengthscales indicates we should partion the power - spectrum so that for @xmath101 the spectrum is isotropic , where @xmath102 is the isotropy scale , between @xmath103 the spectrum is anisotropic and then below the cutoff scale , @xmath74 , the power spectrum should be set to zero since there should be no modes on scales larger than @xmath104 . for @xmath105 where the spectrum is isotropic we use a power spectrum resembling equation ( [ eq : iostropicps ] ) @xmath106 note the additional factor of @xmath107 in the numerator .
for @xmath108 we write the spectrum as the product @xmath109 .
the spectrum @xmath110 is chosen to be a continuation of the inverse power - law given above while the spectrum in the xy directions , @xmath111 , is the spectrum of the isotropic / homogeneous region in these directions fixed at @xmath112 . the spectrum for @xmath108 is thus @xmath113 the reader may verify the power spectrum defined by equations ( [ eq : psb ] ) and ( [ eq : psa ] ) is normalized .
this anisotropic three - dimensional power spectrum yields a one - dimensional spectrum along the z direction given by @xmath114 for @xmath115 which is exactly the same as the one - dimensional spectrum used in kneller & mauney @xcite . there is no analytic formula for the two - point structure function for _ randomly _ orientated separations using this power spectrum but if we consider the two - point structure function of the random field for points orientated along the z direction then we can compute that in this direction @xmath116 compared to the isotropic spectrum above , this anisotropic spectrum differs in important ways .
first , even if we set @xmath117 we observe that the lack of power in the region @xmath74 means we have to compensate by increasing the structure / decreasing the correlation by the factor @xmath107 .
this increase is the reason for the appearance of the extra factor @xmath107 in equation ( [ eq : psb ] ) .
next , as we increase the ratio @xmath118 , we push more and more of the structure of the field in the xy direction to ever smaller scales reducing even further the correlation of the field at some fixed non - radial separation @xmath0 compared to the isotropic case . as a function of @xmath119 for two points aligned along the @xmath120 direction to the structure function @xmath121 of two points aligned along the @xmath122 direction at @xmath123 .
the three curves in the figure correspond to @xmath124 ( solid ) , @xmath125 ( dashed ) and @xmath126 ( dot - dashed ) . at every @xmath119 we generated @xmath90 realization of the field and the error bar on each point is the standard deviation of @xmath127 .
the structure function @xmath121 was computed using the correlation function given in ( [ eq : zcorrelation ] ) and the relationship @xmath128 .
the inputs to the random field generator were @xmath96 , @xmath97 .
[ fig : grf3daniso ] ] this extra power at small scales can be seen in figure ( [ fig : grf3daniso ] ) which is a plot of the ratio of the one - dimensional two - point structure function in the x direction relative to the structure function along the z direction at the same separation scale for three values of @xmath129 . as promised ,
when @xmath130 there is an equal amount of structure in the field along both radial and non - radial directions but as @xmath129 increases we push more and more of the structure of the field in the xy direction to smaller scales .
the anisotropic power spectrum we have constructed means the turbulence along different parallel rays is less correlated than the turbulence along two rays at the same separation when the power spectrum is isotropic .
if that s the case then the transition probabilities for the neutrinos travelling along those two rays should also be less correlated and below we quantify the decrease .
now that we have the random fields to model the turbulence we are all set to generate turbulent profiles and send neutrinos and antineutrinos through them . to achieve higher efficiency we follow six neutrinos and six antineutrinos simultaneously through every realization of the turbulence with one neutrino and one antineutrino emitted at @xmath131 .
each time we generate a new realization we end up with a different set of transition probabilities so by repeating the calculation many times - in our case a minimum of one thousand times but often much larger - we can create an ensemble of transition probabilities of size @xmath132 from each emission point .
once we have our ensemble we can then go ahead and compute means @xmath133 , variances @xmath134 , and , of course , correlations @xmath135 the correlation of the antineutrino transition probabilities will be denoted as @xmath136 . in the large n
limit the error on the correlation is expected to be @xmath137 . combining the results from the six emission points we can form fifteen separations @xmath0 so fifteen correlations but two points must be remembered : first , groups of them will cluster e.g. we will have a value for the correlation at @xmath138 but also two more at @xmath139 and @xmath140 and second , these groups of transition probability correlations are themselves correlated - half the data in each correlation value is the same for all members of the cluster .
nevertheless these clusters are useful because they serve as a consistency check - we should expect the results to be similar within each cluster - and also they give us an indication if the error in the results are comparable to the expected , large - n error @xmath141 given above .
we also need to specify the neutrino mixing parameters we have used .
the hierarchy will be set to normal and we shall comment on how our results translate to the inverted hierarchy .
as discussed , the neutrino energy will be fixed at @xmath142 , typical of supernova neutrino energies and we shall set the neutrino mixing parameters to be @xmath143ev@xmath144 , @xmath145ev@xmath144 , @xmath146 , and @xmath147
. the recent measurements of the last mixing angle @xmath22 by t2k @xcite , double chooz @xcite , reno @xcite and daya bay @xcite are all in the region of @xmath148 .
we shall adopt this value for the majority of this paper but this result is sufficiently new that we shall show on occasion results with multiple values of @xmath22 in order to put this result in context .
finally , the turbulence amplitude @xmath40 will be allowed to vary but we shall focus upon larger values . with the measurement of a large value of @xmath22 the turbulence effects are negligible for amplitudes of order @xmath149 @xcite . for each of the neutrino emission points @xmath120 . from bottom to top
the emission points are @xmath150 .
the turbulence amplitude is set to @xmath151 , we used @xmath152 , @xmath95 for the 3-d turbulence field generator , and the neutrino mixing parameters used are those given in the text with @xmath153 .
[ fig : f(p23 ) ] ] ( circles ) , @xmath154 ( squares ) and @xmath155 ( triangles ) of the neutrinos emitted at @xmath156 as a function of the parameter @xmath53 keeping the ratio @xmath157 fixed at @xmath158 .
the error bars are not the error on the mean but rather the standard deviation of the samples .
the turbulence amplitude is set to @xmath151 and @xmath153[fig : meanp23vsnk ] ] before we show our results for the correlation of the transition probabilities as a function of the emission separation , we consider first the statistical properties of the ensembles for each emission point .
in addition to being interesting in their own right and useful as a reference , these calculations allow us to test that our 3d random field generator is working properly because the ensembles for each point of emission should be consistent and independent of @xmath120 . in figure ( [ fig : f(p23 ) ] ) we show the frequency distribution of @xmath154 for the six emission locations @xmath120 using the mixing paramaters given above , @xmath159 , @xmath152 , @xmath95 and @xmath160 and the isotropic power spectrum .
the sample size is @xmath161 for each emission point . in each panel of the figure the reader will observe that the transition probability is almost uniformly distributed - there is a slight decrease in the frequency of higher values of @xmath154 - but , more importantly , there is no observed trend with @xmath120
. a closer inspection of figure ( [ fig : f(p23 ) ] ) also hints at some correlation : the bottom few panels of the figure are very similar .
we have reproduced this calculation for other choices of the @xmath53 and @xmath55 paramaters .
the results are shown in figure ( [ fig : meanp23vsnk ] ) where we plot the mean values and standard deviations of @xmath162 , @xmath154 and @xmath155 for ensembles of neutrinos emitted at @xmath156 as a function of the parameter @xmath53 keeping the ratio @xmath79 fixed at @xmath163 .
there is no discernable trend with @xmath49 and we march on confident that setting @xmath152 and @xmath95 does not bias our results .
we now allow the values of @xmath40 and @xmath22 to float and consider both the isotropic and anisotropic power spectrum .
the evolution of the transition probability means as a function of @xmath40 for the two power spectra and two choices of @xmath22 are shown in figure ( [ fig : meansvscstar ] ) .
there are many interesting trends discussed in detail in kneller & mauney @xcite .
large amplitude turbulence works its way through to affect every mixing channel , not just the h resonance channel @xmath154 , as promised so that by @xmath164 we observe @xmath165 , @xmath166 , @xmath167 , and @xmath168 , @xmath169 , @xmath170 . to put this in context , in the absence of turbulence all these transition probabilities are zero when @xmath171 .
the only _ neutrino _ mixing channel with reasonable sensitivity to @xmath22 is the h resonance channel @xmath154 and even then the disparity in @xmath172 at @xmath173 disappears by @xmath174 .
in contrast the _ antineutrinos _ are very sensitive to @xmath22 even at large turbulence amplitudes : the expectation value for @xmath175 varies by a factor of @xmath176 when @xmath22 is changed from @xmath177 to @xmath153 , @xmath178 and @xmath179 on the other hand change by @xmath180 orders of magnitude between the same limits .
while these trends are interesting , the purpose of figure ( [ fig : meansvscstar ] ) is to compare the use of the isotropic and anisotropic power spectra .
except for the h resonance mixing channel @xmath154 , the isotropic power spectrum gives values of @xmath181 which are smaller than the anisotropic spectrum .
the neutrinos are more sensitive to the turbulence when the power spectrum is anisotropic because the neutrinos are sensitive to the amplitude of the turbulence modes of order the neutrino oscillation wavelength @xcite which is typically in the range of @xmath182 in the h resonance region .
the anisotropic spectrum removed all power for the fluctuations in the radial direction at the long wavelengths above @xmath104 - which is of order @xmath183 in our calculation - and to compensate we needed to increase the power on the smaller wavelengths which means and effective increase of their amplitude .
in fact we already know the exact amount the amplitude is effectively increased because we pointed out the @xmath73 factor that appears in the one dimensional power spectrum in the isotropic case compared to the one - dimensional spectrum derived from the anisotropic turbulence .
thankfully , our expectations are confirmed by figure ( [ fig : meansvscstar ] ) because the increase of all the mixing channels except @xmath154 is on the expected scale of @xmath107 .
the isotropy scale paramater @xmath3 , which sets the scale in the radial direction below which the turbulence is isotropic , does not play a role for these point source statistics .
the one - dimensional power spectrum along the radial direction is independent of the isotropy scale @xmath3 which can be seen when comparing equations ( [ eq:1diostropicps ] ) and ( [ eq:1daniostropicps ] ) .
so if the one - dimensional power spectrum is independent of @xmath3 then the effect of switching the power spectrum from isotropic to anisotropic is solely due to the removal of radial long - wavelength fluctuations .
the transition probability @xmath154 behaves slightly differently but is entirely consistent with the understanding of the effects in the other channels . at smaller amplitudes and the smaller value of @xmath22
there is no effect of the power spectrum switch upon @xmath184 because the depolarization limit has been reached . at the larger mixing angle depolarization
has not achieved and switching the power spectrum leads to the effects as seen in @xmath185 and @xmath175 .
the two - flavor depolarization limit is reached for the @xmath153 case when @xmath186 . at around this same turbulence amplitude
there begins the shift to three - flavor depolarization where @xmath187 . whatever the mixing angle used , we see that the mean value @xmath188 as a function of @xmath40 using the anisotropic spectrum begins the transition at smaller @xmath40 than the same calculation using the isotropic spectrum because of the increased amplitude of the small scale fluctuations in the former case .
we now turn to the correlation of the transition probabilities as a function of the distance between the emission points and consider first the case of the isotropic power spectrum .
our result for the correlation of the transition probabilities , except @xmath179 , as a function of the separation @xmath0 at various values of @xmath22 and turbulence amplitudes @xmath40 is shown in figure ( [ fig : rhovscstar ] ) .
@xmath179 is excluded is because it is difficult to calculate its correlation reliably .
what one notices immediately about the results are that @xmath189 , @xmath190 , @xmath191 and @xmath192 all show little sensitivity to either @xmath22 or @xmath40 - which is in contrast to figure ( [ fig : meansvscstar ] ) .
the reason for the lack of sensitivity of these correlations to @xmath22 and @xmath40 is explained by the exponential distributions these transition probabilities possess .
both the turbulence amplitude and the mixing angle simply ` rescale ' the ensemble of transition probabilities and , as equation ( [ eq : rho ] ) shows , this rescaling can not alter the correlation .
one also sees that the correlation of all these transition probabilities is high , @xmath193 , for all separations @xmath194 .
in contrast the correlation of @xmath154 is sensitive to both @xmath22 and @xmath40 .
when @xmath40 is of order @xmath2 the sensitivity to @xmath22 arises because the distributions of @xmath154 at the different mixing angle choices are very different : for @xmath177 the distribution is uniform , for @xmath153 it is strongly skewed to small values of @xmath154 .
as @xmath40 increases the sensitivity disappears because the distributions at each value of @xmath22 become similar : this is the same behavior seen in figure ( [ fig : meansvscstar ] ) .
finally , for @xmath195 the currently preferred value of @xmath22 gives greater correlation at a given seperation than smaller values of @xmath22 .
the correlation @xmath196 is high for @xmath197 , a scale of order the proto - neutron star diameter , at for @xmath195 and decreases rapidly as @xmath40 increases . for @xmath198
the transition probability @xmath154 of two neutrinos emitted from points on the proto - neutron star separated by a distance greater than @xmath199 are essentially independent .
- top panel - @xmath175 - center panel - and @xmath154 - bottom panel - through anisotropic turbulence as a function of the separation between neutrino emission points .
the turbulence amplitude is set at @xmath151 and @xmath153 . in each panel
the correlation of the transition probabilities through the isotropic turbulence is shown as the solid line .
the other curves in each panel correspond to different values of the ratio @xmath200 : @xmath201 are squares joined by long dashed lines , @xmath202 are triangles joined by dash - dot lines , and @xmath203 are diamonds joined by dot double - dash lines .
the error bars on each data point are estimated using the large n limit prediction .
[ fig : rho12,rho13,rho23aniso0.3 ] ] the change to the mean point source transition probabilities when switching to an anisotropic power spectrum is both understandable and measurable but , overall , the effects are small and of the order of factors of @xmath107 i.e. the amplitude by which the small scale fluctuations in the anisotropic spectrum increased in amplitude compared to the isotropic spectrum .
that insensitivity no longer holds when we examine the correlations of the transition probabilities because these quantities are functions of the isotropy scale paramater @xmath3 .
the correlations of the transition probabilities as a function of the separation between the emission points is strongly sensitive to the amount of turbulence power in the perpendicular directions and increasing @xmath3 relative to the fixed scale @xmath4 shifts the power from long wavelength , small @xmath204 and @xmath205 , to much shorter wavelengths , as shown in figure ( [ fig : grf3daniso ] ) .
the effects of introducing and varying @xmath200 are shown in figure ( [ fig : rho12,rho13,rho23aniso0.3 ] ) for the case @xmath206 and @xmath153 . for @xmath201
the difference between the isotropic and anisotropic power spectra are minimal but at larger ratios of the two scales the correlation at some given separation @xmath0 drops noticeably in all three channels though the reduction in the correlation of @xmath185 and @xmath175 is not as severe as that for the transition probability @xmath154 .
it is still the case that the correlation of @xmath185 and @xmath175 at typical proto - neutron star radii of @xmath207 is larger than 0.5 if @xmath208 for this particular mixing angle choice and turbulence amplitude .
pushing even more power to smaller scales would lead to minimal correlation of these two transition probabilities . in the case of @xmath154 ,
the correlation at @xmath207 is already small for this mixing angle and turbulence amplitude even in the isotropic and @xmath201 cases so pushing more power of the fluctuations in perpendicular directions to smaller wavelengths completely removes the correlation of @xmath154 over the proto - neutron star radial scale .
if we now vary the turbulence amplitude we generate figure ( [ fig : rho12,rho13,rho23aniso9 ] ) which is , again , for a mixing angle of @xmath153 . examining the results
we quickly observe the same general trends with changes in @xmath209 as seen in figure ( [ fig : rho12,rho13,rho23aniso0.3 ] ) : increasing @xmath209 reduces the correlation with @xmath196 affected to a greater degree than @xmath189 , @xmath190 , @xmath191 and @xmath192 . likewise
, the trends seen in figure ( [ fig : rhovscstar ] ) for changes in @xmath40 are also reproduced .
supernova turbulence and its effects upon both the flavor composition of neutrinos that pass through it and their correlations depend upon many numerous parameters one needs to introduce to describe the turbulence .
all affect the result and here we try to succinctly summarize our results . for a neutrino energy of @xmath36 and using a supernova density profile taken from a simulation @xmath210 post - bounce , we find in a normal hierarchy that the correlation of the h resonance mixing channel transition probability @xmath154 as a function of the emission separation @xmath0 drops considerably as @xmath40 increases for both the cases of isotropic and anisotropic turbulence .
if the turbulence amplitude is of order @xmath173 then the correlation of the transition probability @xmath154 for neutrinos emitted from opposite sides of the proto - neutron star , i.e. separated by @xmath6 , is marginal for the isotropic spectrum and for the anisotropic only when @xmath211 .
for @xmath198 there is essentially no correlation of the h resonance transition probabilities for neutrinos emitted from opposite sides of the proto - neutron star . at these amplitudes
the turbulence along parallel trajectories separated by @xmath6 is just too different to permit any correlation of this transition probability in the supernova neutrino burst signal .
if we switch to an inverted hierarchy then it will be the transition probability @xmath178 which behaves this way .
in contrast , the correlation of the transition probabilities @xmath185 , @xmath175 , @xmath212 and @xmath178 in a normal hierarchy as a function of emission separation is largely independent of @xmath40 and @xmath22 .
the correlation decreases as the ratio @xmath118 increases but remains significant for separations of order the proto - neutron star radius even for @xmath213 . when switching to an inverted hierarchy the mixing channels which behave this way are @xmath185 , @xmath154 , @xmath212 and @xmath179 .
these mixing channels , particularly @xmath185 and @xmath212 , are the most promising for observing flavor scintillation assuming the energy resolution of our neutrino detectors does not wash out the effect and the temporal correlation remains high .
this work was supported by doe grant de - sc0006417 , the topical collaboration in nuclear science
neutrinos and nucleosynthesis in hot and dense matter , doe grant number de - sc0004786 , and an undergraduate research grant from nc state university .
j. t. pantaleone , phys .
b , * 287 * , 128 ( 1992 ) .
s. samuel , , * 48 * , 1462 ( 1993 ) . h. duan , g. m. fuller and y. z. qian , , 123004 ( 2006 ) s. pastor , g. g. raffelt and d. v. semikoz , , 053011 ( 2002 ) s. hannestad , g. g. raffelt , g. sigl and y. y. y. wong , , 105010 ( 2006 ) [ erratum - ibid .
d * 76 * , 029901 ( 2007 ) ] h. duan , g. m. fuller , j. carlson and y. z. qian , 105014 ( 2006 ) g. g. raffelt and a. y. smirnov , , 081301 ( 2007 ) [ erratum - ibid .
d * 77 * , 029903 ( 2008 ) ] raffelt , g. g. , 125015 ( 2008 ) duan , h. , & friedland , a. , , * 106 * 091101 ( 2011 ) reid , g. , adams , j. , & seunarine , s. , 085023 ( 2011 ) chakraborty , s. , fischer , t. , mirizzi , a. , saviano , n. , & toms , r. , 151101 ( 2011 ) galais , s. , kneller , j. , & volpe , c. , journal of physics g nuclear physics , * 39 * 035201 ( 2012 ) cherry , j. f. , carlson , j. , friedland , a. , fuller , g. m. , & vlasenko , a. , 261104 ( 2012 ) sarikas , s. , tamborra , i. , raffelt , g. , hdepohl , l. , & janka , h .-
, 113007 ( 2012 ) mirizzi , a. , & serpico , p. d. , 231102 ( 2012 ) h. duan , g. m. fuller and y. z. qian , arxiv:1001.2799 [ hep - ph ] . h. duan and j. p. kneller , j. phys .
g * 36 * , 113201 ( 2009 ) [ arxiv:0904.0974 [ astro-ph.he ] ] .
s. p. mikheev and a. i. smirnov , nuovo cimento c * 9 * 17 ( 1986 ) r. c. schirato and g. m. fuller , arxiv : astro - ph/0205390 . k. takahashi , k. sato , h. e. dalhed and j. r. wilson , astropart
. phys . * 20 * 189 ( 2003 ) g. l. fogli , e. lisi , d. montanino and a. mirizzi , 033005 ( 2003 ) r. tomas , m. kachelriess , g. raffelt , a. dighe , h. t. janka and l. scheck , jcap * 0409 * 015 ( 2004 ) s. choubey , n. p. harries and g. g. ross , 053010 ( 2006 ) j. gava , j. p. kneller , c. volpe and g. c. mclaughlin , 071101 ( 2009 ) murphy , j. w. , & meakin , c. , 74 ( 2011 ) dolence , j. c. , burrows , a. , murphy , j. w. , & nordhaus , j. , arxiv:1210.5241 ( 2012 ) ott , c. d. , abdikamalov , e. , moesta , p. , et al . , arxiv:1210.6674 ( 2012 ) hanke , f. , marek , a. , mller , b. , & janka , h .-
, 138 ( 2012 ) pejcha , o. , & thompson , t. a. , 106 ( 2012 ) mller , b. , janka , h .- t . , & heger , a. , 72 ( 2012 ) takiwaki , t. , kotake , k. , & suwa , y. , 98 ( 2012 ) lentz , e. j. , bruenn , s. w. , harris , j. a. , et al . ,
arxiv:1301.1326 ( 2013 ) loreti , f. n. and qian , y .- z .
and fuller , g. m. and balantekin , a. b. , 6664 ( 1995 ) g. fogli , e. lisi , a. mirizzi and d. montanino , jcap * 0606 * 012 ( 2006 ) [ arxiv : hep - ph/0603033 ] a. friedland and a. gruzinov , arxiv : astro - ph/0607244 kneller , j. and volpe , c. , 123004 ( 2010 ) abe , k. , abgrall , n. , ajima , y. , et al .
, 041801 ( 2011 ) abe , y. , aberle , c. , akiri , t. , et al . , 131801 ( 2012 )
ahn , j. k. , chebotaryov , s. , choi , j. h. , et al .
, 191802 ( 2012 ) an , f. p. , bai , j. z. , balantekin , a. b. , et al . , 171803 ( 2012 ) | the transition probabilities describing the evolution of a neutrino with a given energy along some ray through a turbulent supernova are random variates unique to each ray .
if the source of the neutrinos were a point then all neutrinos of a given energy and emitted at the same time which were detected in some far off location would have seen the same turbulent profile therefore their transition probabilities would be exactly correlated and would not form a representative sample of the underlying parent transition probability distributions .
but if the source has a finite size then the profiles seen by neutrinos emitted from different points at the source will have seen different turbulence and the correlation of the transition probabilities will be reduced . in this paper
we study the correlation of the neutrino transition probabilities through turbulent supernova profiles as a function of the separation @xmath0 between the emission points using an isotropic and an anisotropic power spectrum for the random field used to model the turbulence .
we find that if we use an isotropic power spectrum for the random field , the correlation of the high ( h ) density resonance mixing channel transition probability is significant , greater than 0.5 , for emission separations of @xmath1 , typical of proto neutron star radii , only when the turbulence amplitude is less than @xmath2 ; at larger amplitudes the correlation in this channel drops close to zero for this same separation of @xmath1 . in contrast , there is significant correlation in the low ( l ) density resonant and non - resonant channels even for turbulence amplitudes as high as 50% . switching to anisotropic spectra requires the introduction of an ` isotropy ' parameter @xmath3 whose inverse defines the scale below which the field is isotropic .
we find the correlation of all transition probabilities , especially the h resonance channel , strongly depends upon the choice of @xmath3 relative to the long wavelength radial cutoff @xmath4 .
the spectral features in the h resonance mixing channel of the next galactic supernova neutrino burst may be strongly obscured by large amplitude turbulence when it enters the signal due to the finite size of the source while the presence of features in the l and non resonant mixing channels may persist , the exact amount depending upon the degree of anisotropy of the turbulence . |
much work has been devoted to realize topologically nontrivial states of matter with topologically protected surface states .
majorana surface states are expected to be formed at the boundaries and vortex cores of topological superconductors ( tss).@xcite such states have been expected in a one - dimensional ( 1d ) quantum wire with spin - orbit interaction ( soi ) placed under a zeeman field and in proximity to a bulk superconductor,@xcite in cold atoms in optical lattices with effective gauge fields generated by spatially varying laser fields,@xcite bilayer electron gases in semiconductor heterostructure with interlayer coulomb coupling,@xcite among others .
several groups of experimentalists have recently reported that they have observed the signatures of majorana fermions appearing at the ends of nanowires attached to superconductors.@xcite for a review we refer to ref . .
the effect of spatial inhomogeneity on a superconducting quantum wire has been a nontrivial problem.@xcite the energy distributions of the end states have been obtained for the dirac equation with random mass and a 1d spinless superconductor.@xcite the interplay of disorder and correlation in 1d tss has also been investigated.@xcite among the possible realizations of spatial inhomogeneity , quasiperiodic ( harper ) potential modulation forms a special class in that in 1d system all the single particle eigenstates become localized at the same modulation strength .
quasiperiodic potentials have been experimentally studied in optically trapped cold atom systems @xcite as well as in solid state , misfit compound @xcite systems .
signatures of a hofstadter butterfly - like band structure have been observed in a van der waals system of monolayer graphene on top of a hexagonal boron nitrade surface.@xcite the 1d effectively spinless superconductor , expected in 1d superconductor with strong spin - momentum coupling under magnetic field , is of the d symmetry class.@xcite therefore its topology is classified by a @xmath0 topological number .
this suggests that the boundary between two topologically non - trivial 1d superconductors would not have a localized mode .
boundary phenomena between topologically equivalent or distinct phases , with harper or fibonacci potentials , have been experimentally studied using photonic quasicrystals.@xcite in the scenario of refs .
the chemical potential needs to lie close to the band edge so that the band degeneracy is removed by the external magnetic field .
however , the present authors have observed that , by a quasiperiodic lattice modulation with a fixed wavenumber , effectively single - band superconductor with end majorana fermions is realized even when the chemical potential is closer to the center of the original cosine band , because energy separations are introduced within each of the zeeman - split bands.@xcite we have also demonstrated that this physics is stable even in the presence of a hubbard - like on - site interaction and/or a harmonic trapping potential .
more recently , the effect of incommensurate potentials on 1d p - wave superconductors have been studied.@xcite _ commensurate _ diagonal or off - diagonal harper model has also attracted theoretical attention.@xcite here we are interested in characterizing the new ts regions further , focusing on when they emerge , and what happens when the quasiperiodically modulated quantum wire is connected with other wires with different modulation phases or an unmodulated one . particularly , we would like to ( i ) clarify the correspondence between the energy spectrum of the single particle states and emergence of the ts states , in the presence of either a quasiperiodic modulation with a general wavenumber or the external magnetic field to a general direction , ( ii ) understand the effect of phase jumps of the quasiperiodic modulation and that of a quasiperiodic modulation applied to only a limited part of the one - dimensional system , and ( iii ) study the effect of a quasiperiodic modulation of the _ hopping _ parameter . in section
ii we define our model and introduce our bogoliubov - de gennes ( bdg ) equation approach to study the system . we find that with a lattice - site energy modulation with a generic wavenumber , quasiperiodic or periodic , new topological superconducting regions with end majorana fermions emerge .
we also study the effect of the direction of the external magnetic field . then in section iii we observe that those regions are stable against phase jumps in one - dimensional systems , while when the modulated wire is connected with an unmodulated wire , the locations of the localized modes are determined by which of the wires becomes ts .
we also study the case with quasiperiodic modulation in the intersite hopping parameter , and find that multiple topological transitions into and out of ts states occur .
finally , in section iv , we summarize our findings .
we consider a one - dimensional quantum wire parallel to the @xmath1 direction , coupled to a bulk superfluid whose surface is perpendicular to @xmath2 .
we study a tight - binding one - dimensional model of spin-@xmath3 fermions with the rashba - type spin - orbit coupling , the mean - field coupling to the bulk superconductor , and the zeeman energy due to the external magnetic field @xmath4 .
the hamiltonian we have adopted is @xmath5 with @xmath6 .
we set @xmath7 as the unit of energy and set @xmath8 in the following . in the case of @xmath9
the third line becomes @xmath10 here , @xmath11 annihilates a fermion with spin @xmath12 at site @xmath13 , @xmath14 , @xmath15 determines the nearest - neighbor hopping , @xmath16 is the rashba - type soi , @xmath17 is the coupling to the bulk superconductor , @xmath18 is the zeeman energy , @xmath19 is the chemical potential , and @xmath20 is the site energy for spin @xmath21 on site @xmath22 . in this work
we limit our discussion to the case with @xmath23 . in the following
we introduce a quasiperiodic modulation to the site energy , and study the energy distribution of the single particle state and its correspondence with the realization of majorana end modes . ) for @xmath24 , @xmath25 , @xmath26 , @xmath9 .
the single particle hamiltonian , obtained by setting @xmath27 and @xmath28 in eq .
( [ eqn : hamiltonian_gen ] ) , is diagonalized for @xmath29 ( @xmath30 ) and each eigenstate is plotted in a color corresponding to the expectation value of the @xmath31 component of the spin for that eigenstate . the plot for @xmath32 is similar to that for @xmath33 ( not shown ) except that the former is scaled horizontally , because of the duality of the lattice between @xmath34 and @xmath35 .
@xmath36 is the self - similar point , at which the modulation potential equals the unmodulated band width for @xmath37 . , width=327 ] let us first consider the case of @xmath9 . for an infinitely long system having @xmath38
we easily obtain the single particle energy as a function of the quasimomentum @xmath39 , @xmath40 we call them the upper and lower rashba
zeeman ( rz ) bands .
@xcite the mapping of the hamiltonian to that of a spinless system is possible if @xmath19 lies in only one of the rz bands .
@xcite in such a case , by introducing the pairing @xmath17 such that @xmath41 , we obtain the topological superconductor phase .
we consider a site potential which is given by @xmath42 in which @xmath43 , @xmath44 , @xmath45 is the phase of the potential at the center of the system and @xmath46 , in which @xmath47 is a real number such that @xmath48 .
the potential is periodic for a rational @xmath47 , while it is quasiperiodic for an irrational @xmath47 . in the following we choose
@xmath49 , except when we study the effect of @xmath45 and when we study the effect of phase jumps in the system . in a finite length system with @xmath50 lattice sites
, we numerically obtain the set of single particle level energies . for @xmath51 , each single body wavefunction
is extended for @xmath52 and localized for @xmath53 in the @xmath54 limit .
@xcite the self - similar structure of the two - dimensional spectrum plotted against @xmath55 for @xmath56 is called the hofstadter butterfly .
@xcite when the @xmath57 energy levels obtained are plotted against various values of @xmath55 , for @xmath58 , the spectrum shows a self - similar structure resembling two hofstadter butterflies shifted in energy and braided together , as shown in fig .
[ fig : doublehofstadter ] .
we call this structure the double hofstadter butterfly .
@xcite the spin - orbit coupling @xmath16 mixes the spin - up states and spin - down states differently at each value of the quasimomentum @xmath39 of the resulting rz bands .
therefore the spin - independent site potential , which has components with @xmath59 and @xmath60 , further mixes the upper and lower rz bands .
most of the states in the resulting double hofstadter butterfly do not have a completely polarized spin .
we may , however , obtain the expectation value of the @xmath31 component of the spin , @xmath61 , for each of the @xmath57 eigenstate .
the energy levels plotted in fig . [ fig : doublehofstadter ] have been color - coded according to the value of @xmath61 .
we find that , for a fixed value of @xmath55 , the sets of states from two hofstadter - butterfly - like structures with separated values of spin polarizations overlap within some energy ranges . in some regions in energy
there are no single particle states , even inside the range of @xmath62 , which was occupied by states of eq .
( [ eqn : rzbands ] ) before the introduction of the site potential modulation eq .
( [ eqn : sitepotential ] ) .
other regions are occupied by states in only one of the hofstadter - butterfly - like structure .
we study the consequences of the site potential modulation on the realization of ts states for the many - body states with a finite chemical potential in the following .
the hamiltonian ( [ eqn : hamiltonian_gen ] ) is bilinear in operators @xmath63 and @xmath64 .
it can be exactly diagonalized in the nambu spinor space @xmath65 , with the basis obtained as the set of the eigenvectors of the bogoliubov - de gennes ( bdg ) equation .
alternatively , fermion pairing via a short - range attractive interaction can also be simulated by introducing a pairing constant @xmath47 and solving the bogoliubov - de gennes equation self - consistently . in a lattice system
the energy cutoff which renormalizes @xmath47 can be in principle determined from the lattice constant .
however , here we fix the value of @xmath17 , a homogeneous proximity pairing , by hand .
this is because we do not expect that the bdg approximation , which simulates the pair formation _ within _
the 1d wire , directly corresponds to our hamiltonian ( [ eqn : hamiltonian_gen ] ) .
we solve @xcite @xmath66 in which @xmath67 is a @xmath68-dimensional vector and @xmath69 are the single - particle components of the hamiltonian , @xmath70 @xmath71 in which @xmath72 . in the following we call the @xmath68-dimensional matrix in the left hand side of eq .
( [ eqn : bdg ] ) @xmath73 .
we work in the limit of low temperature @xmath74 .
we obtain the particle distribution by @xmath75 , \label{eqn : density}\end{aligned}\ ] ] in which @xmath76 ( @xmath74 ) is the fermi distribution function , with @xmath77 being the step function .
the total number of fermions with spin @xmath21 is obtained as @xmath78 . the sum in eq .
( [ eqn : density ] ) is taken over all @xmath79 .
the matrix @xmath73 is hermitian , and has pairs of positive and negative eigenvalues with equal absolute values , because if @xmath80 , we have @xmath81 for a positive ( negative ) eigenvalue , only @xmath82 ( @xmath83 ) contributes to the particle distribution in the @xmath84 limit . ) and smallest positive ( @xmath85 ) eigenvalues of the bdg equation ( [ eqn : bdg ] ) plotted against the chemical potential @xmath19 for @xmath86 , @xmath87 and @xmath88(a ) @xmath89 , ( b ) @xmath90 .
( c e ) amplitudes of the particle- and hole - like parts of the eigenvectors @xmath91 , @xmath92 plotted against the lattice site for @xmath87 and @xmath93(c ) @xmath94 , ( d ) @xmath95 , ( e ) @xmath96.,width=327 ] for @xmath50 sites in the system , because of the spin and particle hole degrees of freedom , we have @xmath68 eigenstates of the bdg equation ( [ eqn : bdg ] ) .
the introduction of @xmath17 opens a gap in the eigenvalue spectrum of eq .
( [ eqn : bdg ] ) in the absence of the spin - orbit coupling @xmath16 or the zeeman field @xmath18 . with the spin - orbit coupling and the zeeman field , when the chemical potential @xmath19 satisfies @xmath97 so that the kitaev model @xcite is effectively realized in the spinful case , the system is a topological superconductor for @xmath98 when @xmath4 is in the @xmath31@xmath99 plane.@xcite in this case the bdg equation has two eigenstates with @xmath100 .
let us consider the @xmath101-th and @xmath102-th smallest eigenvalues , @xmath103 and @xmath104 , which satisfy @xmath105 because eigenvalues appear in pairs with same absolute value and opposite signs as mentioned above .
we can only have an even number of vanishing eigenvalues , and if we have them @xmath103 and @xmath104 should be included in them ; otherwise @xmath106 . for @xmath107 ,
the number of fermions in the system is negligible
. as @xmath19 is increased , @xmath108 increases , with @xmath109 for @xmath110 .
@xmath104 initially decreases linearly in @xmath19 , reflecting the linear decrease of the required energy to add a single particle in the system .
when it is closer to zero , however , the value of @xmath111 approaches more slowly to zero , especially for a smaller system . when we fit the decrease by a function of the shape @xmath112 , the exponent @xmath113 is roughly in proportion to the system size @xmath50 .
this also suggests that the modes corresponding to @xmath111 are spatially localized . from the eigenvector of @xmath73 corresponding to the eigenvalue @xmath104 , @xmath114
, we define the averaged separation from the system center @xmath115 if the modes localize to the system ends , @xmath116 are close to @xmath117 , the maximum value it can take .
note that , when @xmath103 and @xmath104 are numerically degenerate ( @xmath118 in our work ) , any linear combination of the two eigenvectors that correspond to these eigenvalues would be obtained as the eigenvector corresponding to @xmath104 , so within our bdg calculation we do not directly observe eigenmodes that are localized at only one of the ends of the system . in the density - matrix renormalization group ( dmrg ) simulations of the same model ,
@xcite however , a pair of majorana modes localized at each end of the system have been obtained .
we believe that the dmrg calculation , with limitation on the entanglement entropy between the two ends of the system , automatically chooses less entangled degenerate ground states , which are connected to each other by operating either of the two majorana operators , for the subspaces with even and odd numbers of fermions .
we find that the localization of @xmath119 and @xmath120 , in terms of @xmath121 , precisely corresponds to the localization of the majorana operators observed by dmrg @xcite .
we have plotted the values of @xmath122 as a function of @xmath19 along with the single particle state energy in fig .
[ fig : localization ] ( a b ) . in fig .
[ fig : localization ] ( a ) with @xmath123 , the overlapping rz bands are clearly observed , and @xmath122 vanish only when the chemical potential crosses only one of the rz bands . in fig . [
fig : localization ] ( b ) with @xmath124 several regions with vanishing @xmath122 are found , each corresponding to the energy region with states from just one of the two hofstadter - butterfly - like structure .
the components of the eigenvector corresponding to the eigenvalue @xmath104 are shown in fig .
[ fig : localization ] ( c e ) for values of @xmath19 approaching to one of the regions with vanishing @xmath122 .
the localization of the mode is clearly observed with an increase of @xmath116 . in all regions with vanishing @xmath122
, we observe a clear localization of the corresponding eigenvectors of @xmath73 . ) color - coded according to the value of @xmath125 for @xmath126 , @xmath127 , @xmath128 and @xmath129 , @xmath130 .
lower figures : grayscale plots of @xmath104 for the same parameters with @xmath131.,width=327 ] ) color - coded according to the value of @xmath125 for @xmath126 , @xmath127 , @xmath132 and @xmath129 , @xmath130 .
lower figures : grayscale plots of @xmath104 for the same parameters with @xmath131.,width=327 ] in ref .
@xcite we fixed the wave vector @xmath55 of the quasiperiodic lattice modulation .
however , as we have observed in fig .
[ fig : doublehofstadter ] , we have effectively single - band regions of the chemical potential for a wide range of the value of @xmath55 .
therefore it is interesting to study the dependence of the appearance of topologically non - trivial superfluid with end majorana fermions on the value of @xmath55 .
especially it is intriguing whether a @xmath55 such that @xmath133 is an integer has any difference . in figs .
[ fig : l400lambda0.2 ] and [ fig : l400lambda0.5 ] we plot the value of @xmath104 in a color code , along with the single particle state energies obtained from eq .
( [ eqn : hamiltonian_gen ] ) . for smaller values of @xmath17
, we notice that vanishing eigenvalue corresponds to regions covered by a single band of single particle eigenstates , regardless of the value of @xmath55 .
particularly , even for values of @xmath55 such as @xmath134 and @xmath135 , @xmath104 vanishes when the chemical potential lies in the region where single particle eigenstates are effectively spinless . in these regions , because of the effectively single band structure of the non - interacting hamiltonian , the kitaev model is realized , with majorana fermion modes localizing at system ends .
@xcite the value of @xmath104 becomes closer to @xmath89 as @xmath50 is increased , reflecting that the end modes are more separated . for smaller @xmath50 ,
the separation between the single particle eigenstates is comparable to @xmath17 when @xmath136 . in this case ,
strength of the induced superconductivity depends on the relative location of a level and the chemical potential . when the separation is larger ( the density of states is lower ) the value of @xmath104 usually stays larger , but changes rapidly .
@xmath104 becomes more homogeneous and generally reduced as @xmath17 is increased inside each sub - band which does not overlap in energy with another .
we note that , while @xmath137 is always larger than @xmath138 for @xmath110 , there are sub - bands having fermions almost polarized in the @xmath139 direction ( @xmath140 ) even when @xmath141 with less than half filling factor ( @xmath142 ) .
remarkably , even when the chemical potential lies in one of such bands , a ts state with end majorana fermions can be formed . however , as @xmath17 is increased , smaller features in the single particle state distribution , which remained in the structure of the value of @xmath104 for smaller @xmath17 , gradually disappear and only the widest single - band regions remain visible . similar simplification of the phase diagram is also observed in fig . 2 of ref . for a quasiperiodic system of spinless fermions .
finally , for @xmath143 , the in - gap state disappears and the eigenvalue spectrum of @xmath73 has a gap of the order of @xmath17 , regardless of the value of @xmath144 , so that the system is topologically trivial . in summary , the new topologically non - trivial regions with end majorana fermions appear for general site energy modulations of the type of eq .
( [ eqn : sitepotential ] ) , and their existence is not limited to some special irrational values of @xmath145 .
the commensurate case can be considered as a kind of multi - band wire .
the possibility of ts states with end majorana modes has also been studied in multi - band systems.@xcite we have observed that while the range of the chemical potential reflects the single particle eigenstate spectrum strongly , the value of @xmath17 also plays an important role .
if @xmath17 is too small @xmath122 vanishes only for regions with higher density of states .
if @xmath17 is too large , smaller features in the single particle spectrum become smeared .
this occurs before @xmath17 exceeds @xmath18 so that the system becomes topologically trivial regardless of the values of @xmath19 . for @xmath146 , @xmath147 , @xmath148 , @xmath149 , and @xmath129 , @xmath150 .
, width=327 ] the existence of the majorana end fermions depends on the direction of the applied magnetic field .
@xcite namely , the effective magnetic field introduced by the rashba spin - momentum coupling needs to have perpendicular component to the external magnetic field . in our model
the rashba spin - momentum coupling is in the @xmath151 direction , so the perpendicular directions lie in the @xmath31@xmath99 plane , here we ask : can the spin - insensitive quasiperiodic modulation change this situation ? to answer this question we now consider @xmath152 , with @xmath153 .
we plot the value of @xmath104 in a color - coded plot in fig .
[ fig : rotatetheta ] .
while the region with vanishing @xmath104 persists up to @xmath154 , which corresponds to @xmath155 , for larger @xmath156 with @xmath157 we no longer have a vanishing eigenvalue of the bdg equation and the system is topologically nontrivial .
therefore , the introduction of the site level modulation still does not lift the limitation in the direction of @xmath4 for ts states to be observed . because @xmath158 still does not exceed @xmath159
, the result above indicates that the @xmath151 component of the applied magnetic field is rather detrimental for the realization of a ts in our model . )
color - coded according to the value of @xmath125 for @xmath160 , @xmath126 , @xmath147 , @xmath132 and @xmath129 , @xmath161 .
lower figures : grayscale plots of @xmath104 for the same set of parameters except that @xmath158 is introduced and @xmath162.,width=327 ] it has been known ( see _ e.g. _ @xcite ) for systems with @xmath37 that when we fix the value of @xmath55 one or more in - gap states emerge within the energy gaps due to the potential @xmath163 .
such in - gap states also exist in our model with non - zero @xmath18 and @xmath16 , and can cross each other or another sub - band because the rz bands are shifted in energy . as we change the value of the phase of the site energy modulation , @xmath45 , the energies of these in - gap states change rapidly , while other states in ` bulk ' sub - bands do not change significantly . in fig .
[ fig : f - l200lambda0.5 ] we change the value of @xmath45 to study the effect of such in - gap states .
we have plotted the single particle state energies color - coded by the value of @xmath164 as well as the value of @xmath104 for different system sizes .
the plot at the bottom with @xmath24 looks similar to that of the single particle state energies at the top , except that the regions with two overlapping sub - bands do not have a vanishing @xmath104 .
we observe that the dependence of the eigenvalues on the choice of the phase @xmath45 becomes weaker as @xmath50 increases .
bulk sub - bands are not shifted , and crossing with a single in - gap state does not usually break the effectively single band situation .
therefore the topological equivalence between systems with different modulation phases is clear . in the next section
we study what happens if the modulation phase abruptly changes inside the quantum wire , or if the modulation disappears from a part of the wire .
in applications to the quantum information field , namely quantum computation utilizing the pair annihilation or creation of multiple majorana fermions via gates , @xcite end majorana fermions should be stable against minor changes of the condition of the internals of the one - dimensional system , which would occur when two or more quantum wires are joined via gates .
our model of modulated lattices is characterized by the pair of the wavenumber @xmath55 and the phase @xmath45 . here , it is of much interest what happens if we have phase jumps of the lattice modulation , which would correspond to joints between quantum wires , in our system .
let us consider a system with @xmath165 phase jumps , @xmath166 in which @xmath167 denotes the largest integer not exceeding @xmath99 and @xmath168 is the distance between phase jumps of @xmath169 .
if the phase jumps do not affect the majorana modes , the regions of @xmath19 with vanishing @xmath122 should not change , and when they vanish , the corresponding eigenmodes should occupy the ends of the system in spite of the internal phase jumps . for @xmath160 , @xmath126 , @xmath147 , @xmath132 , @xmath158 , @xmath129 , @xmath165 and @xmath170 .
( b ) value of @xmath104 plotted against @xmath19 for @xmath171 and @xmath172 .
eigenstates for @xmath173 are plotted for ( c ) @xmath174 , ( d ) @xmath175 , and ( e ) @xmath176 .
, width=327 ] in fig .
[ fig : phij ] we plot the value of @xmath104 for ( a ) a single jump with different sizes of phase jump @xmath169 , and ( b ) different numbers of phase jumps with @xmath172 . in fig .
[ fig : phij ] ( a ) , @xmath177 corresponds to a system without a phase jump .
introduction of a single phase jump almost does not change the locations of the regions with vanishing @xmath122 , though we observe a few curves of kinks in @xmath104 corresponding to a single particle state running between rz bands as @xmath169 is changed . increasing the number of phase jumps does not change the picture significantly , as long as the localization of the end modes is almost within a single section between phase jumps .
this is observed in fig .
[ fig : phij ] ( b ) ; while the values of @xmath104 sometimes differ between systems with different numbers of phase jumps of @xmath178 , the regions with @xmath179 is not shifted or removed . also we find in fig .
[ fig : phij ] ( c e ) that the eigenvalues of @xmath73 other than @xmath122 do not vanish in this case .
there are only one pair of majorana fermions appearing at the both ends of the system , rather than more than one pairs of them appearing also at phase jumps .
the results above reflect the fact that a phase jump does not change the topological character of the system .
if both sides of the introduced phase jump are in the state characterized by the same topological quantum number , boundary states do not form at the phase jump . plotted against @xmath19 for @xmath180 and @xmath160 , @xmath126 , @xmath147 , @xmath181 , @xmath158 .
eigenstates for @xmath182 are plotted for ( b ) @xmath183 ( only the right side of the system is single band ) , ( c ) @xmath184 ( both sides are single band ) , and ( d ) @xmath185 ( only the left side is single band ) . ,
width=327 ] the view above is further confirmed when we remove the lattice modulation from some part of the system , by having @xmath186 in fig .
[ fig : partialmodulation ] ( a ) we plot the value of @xmath104 for different values of @xmath187 for the case in which the part of the system to the left of site @xmath187 has the same parameter as in fig .
[ fig : localization ] ( a ) , whereas the right part has the same parameter as in fig .
[ fig : localization ] ( b ) .
for @xmath188 we observe that @xmath104 vanishes when it vanishes either in fig .
[ fig : localization ] ( a ) with @xmath123 or ( b ) with @xmath124 . in fig .
[ fig : partialmodulation ] ( b d ) we have plotted the spatial distribution of eigenvectors of ( [ eqn : bdg ] ) corresponding to @xmath104 . in fig . [
fig : partialmodulation ] ( b ) with @xmath183 , only the right side of the system with @xmath124 is an effectively single band system at the chemical potential and becomes a topological superconductor .
the eigenvector has most of its amplitude localized equally to the two ends of this region , and the value of @xmath116 is close to @xmath189 , as expected in such a case . also , in fig .
[ fig : partialmodulation ] ( d ) with @xmath185 , only the left side of the system without the lattice modulation becomes a topological superconductor , and the eigenvector is this time localized to the two ends of the left part , with the value of @xmath116 similar to that in ( b ) . on the other hand , in fig .
[ fig : partialmodulation ] ( c ) both sides are topologically nontrivial , and the eigenvector has its amplitudes localized at the two ends of the whole system . as we enlarge the part without lattice modulation by increasing the value of @xmath187 in fig .
[ fig : partialmodulation ] ( a ) , the plot of @xmath104 approaches that in fig .
[ fig : localization ] ( a ) .
this is because the part with the lattice modulation can not form a well - defined topological superconductor if it is too short . in summary ,
when two ts regions are joined , the resulting system become a ts with end majorana states at the ends . on the other hand , if a ts is joined with a topologically trivial chain , the resulting system becomes a ts whose majorana states appear close to where they had been in the original ts .
this does not depend on which of the chains has spatial modulation .
if the chemical potential can be controlled in the system one can control which region has boundary majorana modes in the setting above . , @xmath147 , @xmath190 and @xmath129 , @xmath191 .
bottom : @xmath104 calculated for the same parameters and @xmath158 .
, width=327 ] the correspondence between the ` diagonal ' modulation in the lattice site energy and the ` off - diagonal ' one in the hopping amplitude in quasiperiodic systems has attracted a renewed attention .
@xcite here we study the effect of such a modulation in the hopping amplitude in our system , in which case the hamiltonian is given by @xmath192 with @xmath193\ ] ] and @xmath194 .
in fig . [
fig : hopmod05 ] we plot the single particle state energy as well as the value of @xmath104 obtained by solving the bdg equation similar to ( [ eqn : bdg ] ) but the diagonal ( spin - preserving ) blocks of the kinetic term substituted by the one with the modulated hopping parameter .
we observe that , while the details of the single particle spectrum is changed , the correspondence between the effectively single - band region in the spectrum and the zeros of the @xmath104 is also observed here .
the symmetry of the hamiltonian is not changed by going from the site level modulation to the hopping parameter modulation .
therefore similar response to phase jumps or partial modulations inside the system is expected for the latter case as in the former case , which has been studied in the above .
in summary , we have studied the effect of different types of spatial modulations on the realization of a topological superconductor in a 1d conductor with proximity - induced superfluidity and spin - orbit coupling , extending our previous work , ref . .
the combination of a quasiperiodic site energy modulation with the external zeeman field and the spin - orbit coupling results in the single particle state energy distribution having a fractal pattern , _ double hofstadter butterfly_. within the mean - field , bogoliubov - de gennes approximation , our model hamiltonian can be diagonalized .
we have demonstrated that the smallest positive eigenvalue of the hamiltonian is strongly governed by the single particle energy spectrum for relatively weak induced superfluidity .
localized end modes , which are majorana fermions , exist when two eigenvalues are degenerate at zero energy .
as we change the chemical potential or the modulation wavenumber , we observe multiple reentrant transitions into and out of topologically nontrivial states .
however , for stronger superfluidity , small patterns of the double hofstadter butterfly are smeared from the eigenvalue plot showing the topologically nontrivial parameter ranges .
the resulting topological superconductor is sensitive to the direction of the magnetic field , while the phase of the modulation does not affect the system .
we have also studied the effects of the phase jump of the quasiperiodic potential and what happens when the potential is applied to only a part of the quantum wire .
the results reflect that the system is characterized by a @xmath0 quantum number , that is , all the topologically nontrivial states are indistinguishable , and if two regions with such states are joined , the majorana end modes appear only at the ends of the resulting system . if the chemical potential can be changed , the locations of the end modes can be manipulated .
a quasiperiodic hopping modulation also exhibits a similar phase diagram with reentrant topological transitions .
recently a scheme for topological superconductivity without proximity effect has been proposed @xcite .
our study of the correlation between the band structure and realization of topological superconductivity could also be relevant in such cases .
m. t. acknowledges the hospitality of cavendish laboratory , university of cambridge , where this work was completed .
this work was partially supported by the grant - in - aid for the global coe program `` the next generation of physics , spun from universality and emergence '' from mext of japan .
n. k. is supported by kakenhi ( grants no .
22103005 and no . 25400366 ) and jsps through its first program .
part of the computation in this work has been performed using the facilities of the supercomputer center , institute for solid state physics , university of tokyo and yukawa institute for theoretical physics , kyoto university .
99 n. read and d. green , phys .
b * 61 * , 10267 ( 2000 ) . a. yu . kitaev , phys .- usp . * 44 * , 131 , 268 ( 2001 )
. d. a. ivanov , phys .
lett . * 86 * , 268 ( 2001 ) .
l. fu and c. l. kane , phys .
lett . * 100 * , 096407 ( 2008 ) .
s. fujimoto , phys .
b * 77 * , 220501 ( 2008 ) ; m. sato and s. fujimoto , phys .
b * 79 * , 094504 ( 2009 ) .
m. sato , y. takahashi , and s. fujimoto , phys .
lett . * 103 * , 020401 ( 2009 ) .
y. tanaka , t. yokoyama , and n. nagaosa , phys .
lett . * 103 * , 107002 ( 2009 ) ; j. linder , y. tanaka , t. yokoyama , a. sudb , and n. nagaosa , phys .
lett . * 104 * , 067001 ( 2010 ) .
j. d. sau , r. m. lutchyn , s. tewari , and s. das sarma , phys .
* 104 * , 040502 ( 2010 ) . j. alicea , phys .
b * 81 * , 125318 ( 2010 ) .
r. m. lutchyn , j. d. sau , and s. das sarma , phys .
lett . * 105 * , 077001 ( 2010 ) .
y. oreg , g. refael , and f. von oppen , phys .
lett . * 105 * , 177002 ( 2010 ) .
s. nakosai , y. tanaka , n. nagaosa , phys .
. lett . * 108 * , 147003 ( 2012 ) .
v. mourik , k. zuo , s. m. frolov , s. r. plissard , e. p. a. m. bakkers , and l. p. kouwenhoven , science * 336 * , 1003 ( 2012 ) .
j. g. rodrigo , v. crespo , h. suderow , s. vieira and f. guinea , * 109 * , 237003 ( 2012 ) . m. t. deng , c. l. yu , g. y. huang , m. larsson , p. caroff , and h. q. xu , nano lett .
* 12 * , 6414 ( 2012 ) .
l. p. rokhinson , x. liu , and j. k. furdyna , nature phys .
* 8 * , 795 ( 2012 ) .
t. d. stanescu and s. tewari , j. phys . : condens .
matter * 25 * , 233201 ( 2013 ) .
o. motrunich , k. damle , and d. a. huse , phys .
b * 63 * , 224204 ( 2001 ) .
i. a. gruzberg , n. read , and s. vishveshwara , phys .
b * 71 * , 245124 ( 2005 ) .
p. w. brouwer , m. duckheim , a. romito , and f. von oppen , phys .
* 107 * , 196804 ( 2011 ) ; * 84 * , 144526 ( 2011 ) .
r. m. lutchyn , t. d. stanescu , and s. das sarma , * 85 * , 140513(r ) ( 2012 ) .
a. m. lobos , r. m. lutchyn , and s. das sarma , * 109 * , 146403 ( 2012 ) .
d. bagrets and a. altland , * 109 * , 227005 ( 2012 ) ; p.
neven , d. bagrets , and a. altland , new j. phys . *
15 * , 055019 ( 2013 ) .
s. takei , b. m. fregoso , h. y. hui , a. m. lobos , and s. das sarma , * 110 * , 186803 ( 2013 ) .
i. adagideli , m. wimmer , and a. teker , arxiv:1302.2612
. j. d. sau , and s. das sarma , arxiv:1305.0554 .
et al . _ ,
nature * 453 * , 895 ( 2008 ) ; e. lucioni _ et al .
lett . * 106 * , 230403 ( 2011 ) .
v. brouet _
_ , * 87 * , 041106 ( r ) .
et al . _ ,
science * 340 * , 1427 ( 2013 ) .
a. p. schnyder , s. ryu , a. furusaki , and a. w. w. ludwig , * 78 * , 195125 ( 2008 ) ; a. kitaev , arxiv:0901.2686 .
m. verbin , o. zilberberg , y. e. kraus , y. lahini , and y. silberberg , * 110 * , 076403 ( 2013 ) .
m. tezuka and n. kawakami , phys .
b * 85 * , 140508(r ) .
x. cai , l .- j .
lang , s. chen , and y. wang , * 110 * , 176403 ( 2013 ) .
w. degottardi , d. sen , and s. vishveshwara , * 110 * , 146404 ( 2013 ) .
i. i. satija and g. g. naumis , arxiv:1210.5159 .
lang and s. chen , * 86 * , 205135 ( 2012 ) .
s. ganeshan , k. sun , and s. das sarma , * 110 * , 180403 ( 2013 ) .
e. m. stoudenmire , j. alicea , o. a. starykh , and m. p. a. fisher , phys .
b * 84 * , 014503 ( 2011 ) .
m. kohmoto , phys .
lett . * 51 * , 1198 ( 1983 ) ; c. tang and m. kohmoto , phys . rev .
b * 34 * , 2041 ( 1986 ) .
d. r. hofstadter , phys .
b * 14 * , 2239 ( 1976 ) .
m. iskin , phys .
* 85 * , 013622 ( 2012 ) .
liu , l. jiang , h. pu , and h. hu , phys . rev .
a * 85 * , 021603(r ) ( 2012 ) .
a. c. potter and p. a. lee , phys .
105 * , 227003 ( 2010 ) ; phys .
b * 83 * , 094525 ( 2011 ) .
r. m. lutchyn , t. d. stanescu , and s. das sarma , phys .
lett . * 106 * , 127001 ( 2011 ) ; t. d. stanescu , r. m. lutchyn , and s. das sarma , phys .
b * 84 * , 144522 ( 2011 ) .
rieder , g. kells , m. duckheim , d. meidan , and p. w. brouwer , * 86 * , 125423 ( 2012 ) .
f. mei , s. -l .
zhu , z. -m .
zhang , c. h. oh , and n. goldman , phys . rev .
a * 85 * , 013638 ( 2012 ) .
wu , q .- f .
liang , z. wang , and x. hu , arxiv:1302.3947 . y. e. kraus and o. zilberberg , phys .
lett . * 109 * , 116404 ( 2012 ) .
s. thiem and m. schreiber , arxiv:1212.6337 .
a. farrell and t. pereg - barnea , * 87 * , 214517 ( 2013 ) . | the possibility to observe and manipulate majorana fermions as end states of one - dimensional topological superconductors has been actively discussed recently . in a quantum wire with strong spin - orbit coupling placed in proximity to a bulk superconductor
, a topological superconductor has been expected to be realized when the band energy is split by the application of a magnetic field . when a periodic lattice modulation is applied multiple topological superconductor phases
appear in the phase diagram .
some of them occur for higher filling factors compared to the case without the modulation .
we study the effects of phase jumps and argue that the topologically nontrivial state of the whole system is retained even if they are present .
we also study the effect of the spatial modulation in the hopping parameter . |
in the 1970s paul erds and george purdy @xcite posed the question , given a finite set of points @xmath0 , what is the maximum number of similar copies @xmath4 that can be determined by @xmath3 points in the plane ? .
this problem remains open in general .
however , there has been some progress regarding the order of magnitude of this maximum as a function of @xmath3 .
elekes and erds @xcite noted that @xmath8 for any pattern @xmath0 and they also gave a quadratic lower bound for @xmath4 when @xmath9 or when all the coordinates of the points in @xmath0 are algebraic numbers .
they also proved a slightly subquadratic lower bound for all other patterns @xmath0 .
later , laczkovich and ruzsa @xcite characterized precisely those patterns @xmath0 for which @xmath10 . in spite of this
, the coefficient of the quadratic term is not known for any non - trivial pattern ; it is not even known if @xmath11 exists ! apart from being a natural question in discrete geometry , this problem also arose in connection to optimization of algorithms designed to look for patterns among data obtained from scanners , digital cameras , telescopes , etc .
( see @xcite for further references . )
our paper considers the case when @xmath0 is the set of vertices of an isosceles right triangle .
the case when @xmath0 is the set of vertices of an equilateral triangle has been considered in @xcite . to avoid redundancy
, we refer to an isosceles right triangle as an @xmath12 for the remainder of the paper .
we begin with some definitions .
let @xmath1 denote a finite set of points in the plane .
we define @xmath13 to be the number of triplets in @xmath1 that are the vertices of an @xmath12 .
furthermore , let @xmath14 as it was mentioned before , elekes and erds established that @xmath15 and it is implicit from their work that @xmath16 .
the main goal of this paper is to derive improved constants that bound the function @xmath17 .
specifically , in sections [ sec : lower ] and [ sec : upper ] , we prove the following result : @xmath18 we then proceed to determine , in section [ sec : small cases ] , the exact values of @xmath19 when @xmath20 .
several ideas for the proofs of these bounds come from the equivalent bounds for equilateral triangles in @xcite .
we use the following definition . for @xmath21 , let @xmath22 be the @xmath23 counterclockwise rotation of @xmath1 with center @xmath24 .
furthermore , let @xmath25 be the number of isosceles right triangles in @xmath1 such that @xmath24 is the right - angle vertex of the triangle .
if @xmath21 , then @xmath26 can be computed by simply rotating our point set @xmath1 by @xmath23 about @xmath24 and counting the number of points in the intersection other than @xmath24 .
therefore , @xmath27 due to the fact that an @xmath12 has only one right angle , then @xmath28 that is , the sum computes the number of @xmath12s in @xmath1 . from this identity an initial @xmath29 lower
bound can be derived for @xmath30 using the set @xmath31 we now improve this bound .
the following theorem generalizes our method for finding a lower bound .
we denote by @xmath32 the lattice generated by the points @xmath33 and @xmath34 ; furthermore , we refer to points in @xmath32 as _ lattice points_. the next result provides a formula for the leading term of @xmath13 when our points in @xmath1 are lattice points enclosed by a given shape .
this theorem , its proof , and notation , are similar to theorem 2 in @xcite , where the authors obtained a similar result for equilateral triangles in place of @xmath12s . [ integral ]
let k be a compact set with finite perimeter and area 1 .
define @xmath35 as @xmath36 where @xmath37 .
if @xmath38 is a similar copy of @xmath39 intersecting @xmath32 in exactly @xmath3 points , then @xmath40 given a compact set @xmath41 with finite area and perimeter , we have that @xmath42 where @xmath43 is the scaling of @xmath41 by a factor @xmath44 .
therefore , @xmath45 we see that each error term in the sum is bounded by the perimeter of @xmath38 , which is finite by hypothesis .
thus , @xmath46 the last sum is a riemann approximation for the function @xmath47 over the region @xmath48 , thus @xmath49 since @xmath50 it follows that , @xmath51 as a result , @xmath52 the importance of this theorem can be seen immediately .
although our @xmath29 lower bound for @xmath53 was derived by summing the degrees of each point in a square lattice , the same result can be obtained by letting @xmath39 be the square @xmath54 .
it follows that @xmath55 and @xmath56 an improved lower bound will follow provided that we find a set @xmath39 such that the value for the integral in theorem [ integral ] is larger than @xmath29 . we get a larger value for the integral by letting @xmath39 be the circle @xmath57 . in this case @xmath58 and @xmath59 it was conjectured in @xcite that not only does @xmath60 exist , but it is attained by the uniform lattice in the shape of a circle .
( @xmath61 denotes the maximum number of equilateral triangles determined by @xmath3 points in the plane . ) the corresponding conjecture in the case of the isosceles right triangle turns out to be false .
that is , if @xmath62 exists , then it must be strictly greater than @xmath63 .
define @xmath64 to be the translation of @xmath32 by the vector @xmath65 .
the following lemma will help us to improve our lower bound .
[ lem : lambda ] if @xmath66 and @xmath67 or @xmath68 , then @xmath69 observe that @xmath70 first suppose @xmath71 . since @xmath72 ,
then @xmath73 if and only if @xmath74 and @xmath75 .
this can only happen when either both @xmath76 and @xmath77 are half - integers ( i.e. , @xmath78 ) , or both @xmath76 and @xmath77 are integers ( i.e. , @xmath79 ) .
now suppose @xmath80 . in this case
, because both @xmath81 and @xmath82 are half - integers , we conclude that @xmath83 if and only if both @xmath74 and @xmath75 .
once again this occurs if and only if @xmath84 .
recall that if @xmath39 denotes the circle of area 1 , then @xmath85 is the leading term of @xmath86 .
the previous lemma implies that , if we were to adjoin a point @xmath87 to @xmath88 such that @xmath24 has half - integer coordinates and is located near the center of the circle formed by the points of @xmath88 , then @xmath26 will approximately equal @xmath89 .
we obtain the next theorem by further exploiting this idea .
@xmath90 let @xmath39 be the circle of area 1 , @xmath91 , and @xmath92 .
moreover , position @xmath93 so that its points are centered on the circle formed by the points in @xmath94 ( see figure [ fig : centered ] ) .
we let @xmath95 and @xmath96 , where @xmath97 is a constant to be determined .
( gray points ) centered on set @xmath94 ( black points ) , ( b ) plot of the @xmath98 coefficient of @xmath99 as @xmath100 ranges from 0 to 1 . ]
we proceed to maximize the leading coefficient of @xmath101 as @xmath100 varies from 0 to @xmath102 . by lemma [ lem : lambda ] , there can not exist an @xmath103whose right - angle vertex lies in @xmath94 while one @xmath104 vertex lies in @xmath94 and the other lies in @xmath93 . similarly , there can not exist an whose right angle - vertex lies in @xmath93 while one @xmath104 vertex lies in @xmath94 and the other lies in @xmath93 .
therefore , each with vertices in @xmath105 must fall under one of the following four cases : + _ case 1 : all three vertices in @xmath94_. using theorem [ integral ] , it follows that there are @xmath106 s in this case .
since @xmath107 , the number of s in terms of @xmath3 equals @xmath108 + _ case 2 : all three vertices in @xmath93_. by theorem [ integral ] , there are @xmath109 s in this case .
this time @xmath110 and the number of s in terms of @xmath3 equals @xmath111 + _ case 3 : right - angle vertex in @xmath93 , @xmath104 vertices in @xmath94_. the relationship given by lemma [ lem : lambda ] allows us to slightly adapt the proof of theorem [ integral ] in order to compute the number of s in this case .
the integral approximation to the number of s in this case is given by @xmath112 but @xmath113 so @xmath114 expressing this value in terms of @xmath3 gives @xmath115 + _ case 4 : right - angle vertex in @xmath94 , @xmath104 vertices in @xmath93_. as in case 3 , the number of s is given by @xmath116 now recall that @xmath117 .
it follows that @xmath118 if and only if @xmath119 is farther than @xmath120 from the center of @xmath121 .
thus for small enough values of @xmath122 , the region of integration in equation ( [ eqn : integral ] ) is actually @xmath123 , so it does not depend on @xmath124 .
we consider two subcases .
first , if @xmath125 ( i.e. , @xmath126 ) , then @xmath127 the left side of the above inequality is the radius of @xmath123 , meanwhile the right side is the radius of @xmath128 , thus the region of integration where @xmath129 is nonzero equals @xmath123 .
hence , the number of s equals @xmath130 now we consider the case @xmath131 ( i.e. , @xmath132 ) . in this case , @xmath129 is nonzero for all points in @xmath133 .
thus the number of s in this case equals @xmath134 by equation ( [ circle function ] ) , we have that for @xmath135 , @xmath136 therefore , putting all four cases together ( i.e. , expressions ( [ case 1 ] ) , ( [ case 2 ] ) , ( [ case 3 ] ) , and either ( [ case 4a ] ) or ( [ case 4b ] ) ) , we obtain that the @xmath98 coefficient of @xmath137 equals @xmath138 if @xmath139 , or @xmath140 if @xmath141 .
letting @xmath100 vary from 0 to 1 , it turns out that this coefficient is maximized ( see figure [ fig : centered ] ) when @xmath142 ( this corresponds to when the radius of @xmath93 is approximately 18.87% of the radius of @xmath94 ) .
letting @xmath100 equal this value gives @xmath143 as a decimal approximation to the maximum value attained by the @xmath98 coefficient . at this point
, one might be tempted to further increase the quadratic coefficient by placing a third set of lattice points arranged in a circle and centered on the circle formed by @xmath93 .
it turns out that forming such a configuration does not improve the results in the previous theorem .
this is due to lemma [ lem : lambda ] .
more specifically , given our construction from the previous theorem , there is no place to adjoin a point @xmath24 to the center of @xmath105 such that @xmath144 or @xmath145 .
hence , if we were to add the point @xmath24 to the center of @xmath105 , then any new s would have their right - angle vertex located at @xmath24 with one @xmath104 vertex in @xmath94 and the other @xmath104 vertex in @xmath93 .
doing so can produce at most @xmath146 new s ( recall that @xmath147 in our construction ) . on the other hand , adding @xmath24 to the perimeter of @xmath94 , gives us @xmath148 new s.
we now turn our attention to finding an upper bound for @xmath149 .
it is easy to see that @xmath150 , since any pair of points can be the vertices of at most 6 @xmath12s .
our next theorem improves this bound .
the idea is to prove that there exists a point in @xmath1 that does not belong to many @xmath12s .
first , we need the following definition . for every @xmath21 ,
let @xmath151 and @xmath152 be the dilations of @xmath1 , centered at @xmath24 , by a factor of @xmath153 and @xmath154 , respectively ; followed by a @xmath104 counterclockwise rotation with center @xmath24 .
furthermore , let @xmath155 and @xmath156 be the number of isosceles right triangles @xmath157 with @xmath158 such that @xmath157 is ordered in counterclockwise order , and @xmath159 , respectively @xmath160 , is the hypotenuse of the triangle @xmath157 .
much like the case of @xmath161 , @xmath162 and @xmath163 can be computed with the following identities , @xmath164 [ theorem3]for @xmath165 , @xmath166 by induction on @xmath3 . if @xmath167 , then @xmath168 .
now suppose the theorem holds for @xmath169 .
we must show this implies the theorem holds for @xmath170 .
suppose that there is a point @xmath21 such that @xmath171 .
then by induction , @xmath172 the last equality can be verified by considering the three possible residues of @xmath77 when divided by 3 .
hence , our theorem is proved if we can find a point @xmath21 with the desired property .
let @xmath158 be points such that @xmath100 and @xmath173 form the diameter of @xmath1 .
in other words , if @xmath174 , then the distance from @xmath175 to any other point in @xmath1 is less than or equal to the distance from @xmath100 to @xmath173 .
we now prove that either @xmath100 or @xmath173 is a point with the desired property mentioned above .
we begin by analyzing @xmath163 .
we use the same notation from theorem 1 in @xcite . define @xmath176 and @xmath177 .
it follows from our identities that , @xmath178 and @xmath179 .
furthermore , by the inclusion - exclusion principle for finite sets , we have @xmath180 we shall prove by contradiction that @xmath181 .
suppose that there are two points @xmath182 .
this means that there are points @xmath183 such that the triangles @xmath184 are @xmath12s oriented counterclockwise with right angle at either @xmath185 or @xmath186 .
but notice that the line segments @xmath187 and @xmath188 are simply the @xmath189)-counterclockwise rotations of @xmath190 about centers @xmath185 and @xmath186 respectively .
hence , @xmath191 is a parallelogram with two sides having length @xmath190 as shown in figure [ theoremlemma4](a ) .
this is a contradiction since one of the diagonals of the parallelogram is longer than any of it sides .
thus , @xmath181 .
furthermore , @xmath192 and @xmath193 , so @xmath194 and thus @xmath195 this also implies that @xmath196 since we can follow the exact same argument applied to the reflection of @xmath1 about the line @xmath190 .
we now look at @xmath197 and @xmath198 .
first we need the following lemma . for every @xmath199 , at most one of @xmath200 or @xmath201
belongs to @xmath1 .
let @xmath202 and @xmath203 ( see figure [ theoremlemma4](b ) ) .
note that the distance @xmath204 is exactly the distance @xmath190 but scaled by @xmath153 .
this contradicts the fact that @xmath190 is the diameter of @xmath1 .
let us define a graph @xmath205 with vertex set @xmath206 and where @xmath207 is an edge of @xmath205 , ( i.e. , @xmath208 ) if and only if @xmath209 or @xmath210 .
[ inequalitylemma]@xmath211 the left inequality follows from the fact each edge counts an @xmath12 in either @xmath197 or @xmath198 and possibly in both .
however , if @xmath207 is an edge of @xmath205 so that @xmath209 and @xmath212 , then @xmath213 is a square , so this can only happen for at most one edge .
now , let @xmath214 be the number of edges in @xmath215 incident to @xmath185 .
we prove the following lemma .
[ deglemma ] for every @xmath216 , @xmath217 .
suppose @xmath218 , see figure [ fig : lemma56](a ) .
without loss of generality we can assume that @xmath219 .
if @xmath220 , then we conclude that @xmath221 or @xmath222 , because @xmath223 or @xmath224 .
this contradicts the fact that @xmath190 is the diameter of @xmath1 .
similarly , if @xmath225 and @xmath226 are defined as @xmath227 and @xmath228 , then at most one of @xmath225 or @xmath226 can be in @xmath1 . and [ pathlemma ] .
] we still need one more lemma for our proof .
[ pathlemma ] all paths in @xmath205 have length at most 2 .
we prove this lemma by contradiction .
suppose we can have a path of length 3 or more . to assist us ,
let us place our points on a cartesian coordinate system with our diameter @xmath190 relabeled as the points @xmath229 and @xmath230 , furthermore , assume @xmath231 and that the four vertices of the path of length @xmath232 are @xmath233 , @xmath234 , @xmath235 , and @xmath236 .
our aim is to show that the distance between @xmath236 and @xmath235 contradicts that @xmath44 is the diameter of @xmath1 .
now , if paths of length 3 were possible , then the distance between every pair of points in figure [ fig : lemma56](b ) must be less than or equal to @xmath44 . since @xmath237 then @xmath238 .
now let us analyze the square of the distance from @xmath236 to @xmath235 . because @xmath239 , it follows that @xmath240 but @xmath241 , so @xmath242 and thus @xmath243 equality occur if and only if @xmath244 and @xmath245 ;
otherwise , @xmath246 is strictly greater than @xmath44 , contradicting the fact that the diameter of @xmath1 is @xmath44 .
therefore if @xmath247 or @xmath248 then there is no path of length 3 . in the case that @xmath244 and @xmath245 the points @xmath234 and @xmath236 become the same and so do the points @xmath233 and @xmath235 .
thus we are left with a path of length 1 .
it follows from lemmas [ deglemma ] and [ pathlemma ] that all paths of length 2 are disjoint .
in other words , @xmath205 is the union of disjoint paths of length less than or equal to 2 .
let @xmath249 denote the number of paths of length 2 and @xmath250 denote the number of paths of length 1 , then @xmath251 recall from lemma [ inequalitylemma ] that either @xmath252 or @xmath253 .
if @xmath252 , then @xmath254 so @xmath255 moreover , if @xmath256 , then @xmath257 and we get a minor improvement , @xmath258 so @xmath259 .
we are now ready to put everything together . between the two points @xmath100 and @xmath173
, we derived the following bounds : @xmath260 because the degree of a point must take on an integer value , it must be the case that either @xmath100 or @xmath173 satisfies @xmath261 .
in this section we determine the exact values of @xmath7 when @xmath262 .
we begin with @xmath167 .
since @xmath232 points uniquely determine a triangle , and there is an @xmath103with 3 points ( figure [ fig : smallcases](a ) ) , this situation becomes trivial and we therefore conclude that @xmath270 having @xmath278 implies that we must always have more than one @xmath12 in @xmath1 . hence , we can begin with a single @xmath12 and examine the possible ways of adding a point and producing more @xmath12s .
we accomplish this task in figure [ fig : smallcases](a ) .
the 10 numbers in the figure indicate the location of a point , and the total number of @xmath12s after its addition to the set of black dots .
all other locations not labeled with a number do not increase the number of @xmath12s . therefore , except for symmetries , all the possibilities for @xmath1 are shown in figures [ fig : smallcases](b)(e ) .
suppose that within @xmath1 , every @xmath77-point configuration contains at most @xmath250 @xmath12s .
the number of @xmath12s in @xmath1 can then be counted by adding all the @xmath12s in every @xmath77-point subset of @xmath1 . however , in doing so , we end up counting a fixed @xmath12 exactly @xmath284 times .
because @xmath280 we get , @xmath285 notice that @xmath13 can only take on integer values so , @xmath286 now suppose @xmath287 . if @xmath288 for all @xmath281 with @xmath289 , then by lemma [ sumlemma ] , @xmath290 .
otherwise , by lemma [ p=4 ] , @xmath1 must contain one of the 4 sets shown in figures [ fig : smallcases](b)[fig : smallcases](e ) .
the result now follows by examining the possibilities for producing more @xmath12s by placing a fifth point in the 4 distinct sets . in figures [ fig : smallcases](b ) , [ fig : smallcases](c ) , [ fig : smallcases](d ) , and [ fig : smallcases](e ) we accomplish this task . in the same way as we did in lemma [ p=4 ]
, every number in a figure indicates the location of a point , and the total number of @xmath12s after its addition to the set of black dots .
it follows that the maximum value achieved by placing a fifth point is @xmath291 and so @xmath265 .
the point - set that uniquely achieves equality is shown in figure [ fig : smallcases](f ) .
moreover , there is exactly one set @xmath1 with @xmath292 ( shown in figure [ fig : smallcases](g ) ) , and two sets @xmath1 with @xmath293 ( figures [ fig : smallcases](h ) and [ fig : smallcases](i ) ) .
now suppose @xmath294 . if @xmath295 for all @xmath281 with @xmath296 , then by lemma [ sumlemma ] , @xmath297 .
otherwise , @xmath1 must contain one of the sets in figures [ fig : smallcases](f)[fig : smallcases](i ) .
we now check all possibilities for adding more @xmath12s by joining a sixth point to our 4 distinct sets .
this is shown in figures [ fig : smallcases](f)[fig : smallcases](i ) .
it follows that the maximum value achieved is @xmath298 and so @xmath266 .
the point - set that uniquely achieves equality is shown in figure [ fig : smallcases](j ) .
also , except for symmetries , there are exactly 3 sets @xmath1 with @xmath299 ( figures [ fig : smallcases](k)[fig : smallcases](m ) ) and only one set @xmath1 with @xmath300 ( figure [ fig : smallcases](n ) ) .
now suppose @xmath301 . if @xmath302 for all @xmath281 with @xmath303 , then by the lemma [ sumlemma ] , @xmath304 .
otherwise , @xmath1 must contain one of the sets in figures [ fig : smallcases](j)[fig : smallcases](n ) .
we now check all possibilities for adding more @xmath12s by joining a seventh point to our 5 distinct configurations .
we complete this task in figures [ fig : smallcases](j)[fig : smallcases](n ) . because the maximum value achieved is @xmath305
, we deduce that @xmath267 . in this case
, there are exactly two point - sets that achieve 15 @xmath12s .
the proof for the values @xmath306 and @xmath307 follows along the same lines , but there are many more intermediate sets to be considered .
we omit the details .
all optimal sets are shown in figure [ fig : optimal ] .
inspired by our method used to prove exact values of @xmath308 , a computer algorithm was devised to construct the best 1-point extension of a given base set .
this algorithm , together with appropriate heuristic choices for some initial sets , lead to the construction of point sets with many @xmath12s giving us our best lower bounds for @xmath308 when @xmath309 .
these lower bounds are shown in table 1 and the point - sets achieving them in figure [ fig : bestconst ] .
@xmath3 & 10 & 11 & 12 & 13 & 14 & 15 & 16 & 17 + @xmath310 & 35 & 43 & 52 & 64 & 74 & 85 & 97 & 112 + + @xmath3 & 18 & 19 & 20 & 21 & 22 & 23 & 24 & 25 + @xmath310 & 124 & 139 & 156 & 176 & 192 & 210 & 229 & 252 + * acknowledgements .
* we thank virgilio cerna who , as part of the curm mini - grant that supported this project , helped to implement the program that found the best lower bounds for smaller values of @xmath3 .
we also thank an anonymous referee for some useful suggestions and improvements to the presentation .
p. erds and g. purdy . some extremal problems in geometry iii .
6th southeastern conference in combinatorics , graph theory and comp .
( florida atlantic univ . , boca raton , fla . , 1975 ) _ , pp .
congressus numerantium , no .
xiv , utilitas math . ,
winnipeg , man . ,
p. erds and g. purdy .
some extremal problems in geometry iv . _
7th southeastern conference in combinatorics , graph theory and comp .
( louisiana state univ .
, baton rouge , la . , 1976 )
_ , pp . 307322 .
congressus numerantium , no .
xvii , utilitas math . ,
winnipeg , man . | let @xmath0 be a finite set of points in the plane . for any set @xmath1 of points in the plane
, @xmath2 denotes the number of similar copies of @xmath0 contained in @xmath1 . for a fixed @xmath3 , erds and purdy asked to determine the maximum possible value of @xmath2 , denoted by @xmath4 , over all sets @xmath1 of @xmath3 points in the plane .
we consider this problem when @xmath5 is the set of vertices of an isosceles right triangle .
we give exact solutions when @xmath6 , and provide new upper and lower bounds for @xmath7 . |
the field of quantum plasma physics has a long and diverse tradition @xcite , and is becoming of increasing current interest@xcite , motivated by its potential applications in modern technology ( e.g. metallic and semiconductor nanostructures - such as metallic nanoparticles , metal clusters , thin metal films , spintronics , nanotubes , quantum well and quantum dots , nano - plasmonic devices , quantum x - ray free - electron lasers , etc . ) . due to the recent development of ultrafast spectroscopy techniques , it is now possible to monitor the femtosecond dynamics of an electron gas confined in metallic plasmas . in dense quantum plasmas ,
the number densities of degenerate electrons and/or positrons are extremely high , and the plasma particles ( mainly electrons and positrons ) obey fermi - dirac statistics .
the quantum degeneracy effects start playing a significant role when the de broglie thermal wavelength @xmath0 for electrons is similar to or larger than the average inter - electron distance @xmath1 , i.e. when @xcite @xmath2 or , equivalently , the temperature @xmath3 is comparable or lower than the electron fermi temperature @xmath4 , where the electron fermi energy is @xmath5 so that @xmath6 here @xmath7 is the planck constant divided by @xmath8 , @xmath9 is the electron number density , @xmath10 is the rest electron mass , and @xmath11 is the boltzmann constant .
when the temperature approaches the electron fermi temperature @xmath12 , one can show , by using the density matrix formalism @xcite , that the equilibrium electron distribution function changes from the maxwell
boltzmann @xmath13 to the fermi
dirac distribution @xmath14^{-1}$ ] , where @xmath15 is the electron kinetic energy and @xmath16 is the chemical potential . in a dense fermi plasma , the thomas - fermi screening radius reads @xmath17 which is the quantum analogue of the debye - hckel radius . here
the electron fermi speed @xmath18 is the speed of an electron at the fermi surface .
a measure of the importance of collisions in a dense plasma is the quantum coupling parameter , which is the ratio between the interaction energy @xmath19 and the average kinetic energy @xmath20 of electrons , where @xmath21 is the magnitude of the electron charge . for a classical plasma ,
the kinetic energy is @xmath22 and hence we have @xmath23 in the classical case . in a quantum plasma , we have instead @xmath24 , which gives the quantum coupling parameter @xcite @xmath25 ( where we have left out proportionality constants for the sake of clarity ) is analogous to the classical one when @xmath26 .
the various plasma regimes are illustrated in fig .
[ fig : plasma_diagram ] , where the straight lines correspond to i ) the distinction between the classical and quantum plasmas , @xmath27 , ii ) the limit between collisionless and collisional classical plasmas , @xmath28 , and iii ) the limit between collisionless and collisional quantum plasmas , @xmath29 . in fig .
[ fig : plasma_diagram ] , various experimental and naturally occurring plasmas are exemplified .
also indicated in fig .
[ fig : plasma_diagram ] are the kinetic equations used to model the plasma in each regime .
the vlasov and wigner equations are thus used to model collisionless plasma in the classical and quantum limits , respectively , while `` boltzmann '' indicates collisional kinetic models in a classical plasma .
the kinetic models for a collisional quantum plasma is labeled `` wigner ( + coll ) '' in fig .
[ fig : plasma_diagram ] .
we note that a quantum plasma becomes collisionless @xmath30 when the mean distance between electrons is of the order of the bohr radius @xmath31 , i.e. @xmath32a_0\approx 0.38 a_0 $ ] , where @xmath33 .
this corresponds to a number density of the order @xmath34 , which is three orders of magnitude larger than the electron density in a typical metal , for example gold has @xmath35 at room temperature . in fermi - degenerate matters
, there is an effect called the pauli - blocking , which strongly reduces the electron - electron and electron - ion collisional rates .
namely , at moderate temperatures , only electrons within an energy shell of thickness @xmath22 about the fermi surface ( where the electron energy equals @xmath36 ) can undergo collisions . for these electrons ,
the electron - electron collision rate is of the order @xmath37 , and the average collision rate is obtained by multiplying this expression by @xmath38 .
the resulting collision frequency @xmath39 divided by the electron plasma frequency is @xcite @xmath40 hence , @xmath41 when @xmath42 and @xmath43 , which is relevant for metallic electrons .
for example , at room temperature we have @xmath44 , which is much smaller than the typical collisionless frequency of collective interactions , @xmath45 . also ,
the typical collision rate for electron - lattice ( ion ) collisions , @xmath46 , is smaller than @xmath47 by one order of magnitude . therefore
, a collisionless regime seems to be relevant for free electrons in a metal on a time - scale of the order of a femtosecond . in denser plasmas , such as in stellar interiors @xcite and in inertial fusion schemes @xcite ,
the relative effects of collisions decrease even further and leads to increased electron transport and heat conductivity .
@xmath48 plane : iono : ionospheric plasma , space : space plasma , corona : solar corona , discha : typical electric discharge , tok : tokamak / magnetic fusion experiments , icf : inertial confinement fusion , met : metals and metal clusters , jup : jupiter s core , dwarf : white dwarf star . after refs .
@xcite.,width=377 ] more than sixty five years ago , wigner @xcite introduced a phase - space formalism to treat a quantum state of charged particles in a collisionless quantum system .
he introduced a quantum distribution function of the phase - space variables @xmath49 .
the wigner distribution function is defined as @xmath50 where @xmath51 is the number of particles in the system , @xmath52 and @xmath53 are the sets of coordinates and momenta for the particles , @xmath54 is the time , and @xmath55 is the density matrix .
the wigner function is not a probability density in phase space @xmath56 , because it can take negative values .
the wigner distribution function @xmath57 , corresponding to the wave function @xmath58 , can be expressed as @xmath59 which has the property @xmath60 where the asterisk denotes complex conjugate .
the quantum kinetic equation based on the wigner distribution was developed by moyal @xcite , and is now referred to as the wigner - moyal description for treating statistical effects on electron plasma waves in a quantum plasma @xcite .
analytical investigations of collective interactions between an ensemble of degenerate electrons in a dense quantum plasma dates back to early fifties . specifically , klimontovich and silin @xcite and bohm and pines @xcite presented properties of linear electron plasma oscillations ( epos ) in a dense quantum plasma . in the latter , electrons , positrons , and holes
are degenerate , while ions are cold and classical ( typically , in a dense quantum plasma the ion fermi speed is much smaller than the electron fermi speed ) .
accordingly , electrons , positrons and holes have a fermi - dirac distribution function @xcite , contrary to the boltzmann - maxwell distribution function for charged particles in a classical plasma .
the dispersion relation for high - frequency electron plasma oscillations in a dense quantum plasma with fixed ion background reads ( see appendix b ) @xmath61 which was also obtained by bohm and pines @xcite by performing a series of canonical transformations on the hamiltonian of the system of individual electrons , interacting via the electrostatic force . in the zero temperature limit
, we have ( see appendix b ) @xmath62 \log\left| \frac { \frac{\omega}{k}-v_{fe}+\frac{\hbar k}{2 m_e } } { \frac{\omega}{k}+v_{fe}+\frac{\hbar k}{2 m_e } } \right| \right .
\\ \left .
+ \frac{m_e}{\hbar k v_{fe}}\left[v_{fe}^2-\left(\frac{\omega}{k}-\frac{\hbar k}{2 m_e}\right)^2\right ] \log\left| \frac { \frac{\omega}{k}-v_{fe}-\frac{\hbar k}{2 m_e } } { \frac{\omega}{k}+v_{fe}-\frac{\hbar k}{2 m_e } } \right| \right\}\equiv 1+\chi_e=0 , \end{split } \label{wigner_disp}\ ] ] which , in the limit @xmath63 yields @xmath64 where we have assumed that @xmath65 is real and @xmath66 . here
@xmath65 is the wave frequency , @xmath67 is the wave vector , and @xmath68 is the electron plasma frequency , on the other hand , for small wavenumbers up to terms containing @xmath69 , we obtain from ( [ wigner_disp ] ) @xmath70 where @xmath71 , where @xmath72 is the bohr radius .
equation ( 14 ) shows that the wave dispersion arises due to the finite width of the electron wave function in a dense fermi plasma @xcite .
furthermore , in the low phase speed limit , viz .
@xmath73 , the dielectric constant for ion oscillations reads @xmath74 which yields , with @xmath75 , the ion oscillation frequency @xmath76 where @xmath77 is the ion plasma frequency , and @xmath78 is the ion mass , @xmath79 , and @xmath80 .
for @xmath81 , we have from ( 16b ) @xmath82 where @xmath83 is the sound speed .
dispersion properties of electrostatic waves in an unmagnetized dense quantum plasmas with arbitrary electron degeneracy have been presented by maafa @xcite and melrose @xcite . by using the random phase approximation ,
the permittivity of a degenerate collisionless plasma is given in textbooks @xcite .
furthermore , since the pure electromagnetic wave in a non - streaming unmagnetized dense plasma do not accompany density fluctuations , the wave frequency is @xmath84 , where @xmath85 is the speed of light in vacuum .
theoretical studies of quantum statistical properties of dense plasmas in the presence of electromagnetic waves appear in kremp et al .
@xcite and in textbooks @xcite , while quantum parameter regimes are discussed by bonitz @xcite . in a magnetized dense quantum plasma
, one finds that the external magnetic field significantly affects the dynamics of degenerate electrons and positrons , and subsequently there appear new collective phenomena associated with the electron angular momentum spin @xcite , the electron spin magnetic moment @xcite , and quantized landau energy levels @xcite of the fermions in a strong magnetic field .
it turns out that the thermodynamics and kinetics @xcite , as well as the dispersion properties of both electrostatic and electromagnetic waves @xcite in a quantum magnetoplasma are significantly different from those in an unmagnetized quantum plasma .
it was early recognized that the underlying physics of nonlinear quantum - like equations can be better understood by casting those equations in the form of hydrodynamical ( or the euler ) equations , which essentially represent the evolution of quantum particle densities and momenta .
this was elegantly done by madelung @xcite and bohm @xcite by introducing an eikonal representation for the wave function evolution in the non - stationary schrdinger equation .
the derivation of the madelung quantum fluid equations for the pauli equation with the quantum particle angular momentum spin was presented by takabayashi @xcite , bohm et al .
@xcite , janossy and ziegler - naray @xcite , and others @xcite . to incorporate relativistic effects into the quantum fluid formalism ,
takabayashi derived the quantum electron fluid equations for the klein - gordon equation @xcite and for the dirac equation @xcite .
extensions have also been done to fluid descriptions of the weyl equation for massless spin-1/2 particles ( neutrinos ) by bialynicki - birula @xcite .
recently , there has been growing and vibrant interests in investigating new aspects of quantum plasma physics by developing non - relativistic quantum hydrodynamical ( qhd ) equations @xcite .
the latter include the quantum statistical electron pressure and the quantum force involving tunneling of degenerate electrons through the bohm potential @xcite .
the wigner - poisson ( wp ) model has also been used to derive a set of non - relativistic quantum hydrodynamical ( qhd ) equations @xcite for a dense electron plasma , assuming immobile ions .
the qhd equations are composed of the electron continuity , electron momentum and poisson equations .
the quantum force @xcite appears in the non - relativistic electron momentum equation through the pressure term , which requires knowledge of the wigner distribution for a quantum mixture of electron wave functions , each characterized by an occupation probability .
quantum transport models similar to the qhd plasma model have also been used in superfluidity @xcite and superconductivity @xcite , as well as in the study of metal clusters and nanoparticles , where they are referred to as nonstationary thomas - fermi models @xcite .
the electrostatic qhd equations are useful for studying collective interactions ( e.g. different types of waves , instabilities , quantum fluid turbulence and nonlinear structures @xcite in dense quantum plasmas .
the quantum kinetic and qhd equations have also been generalized to include the electromagnetic , ambient magnetic field and electron angular momentum spin effects @xcite .
the latter give rise to high - frequency spin waves , which can be excited by neutrino beams in supernovae @xcite .
furthermore , studies of numerous collective interactions in dense plasmas are relevant in the context of i ) intense laser - solid density plasma experiments @xcite , where one would be exploring new frontiers in high - energy density physics @xcite ; ii ) in the cores of giant planets and the crusts of old stars @xcite ; iii ) superdense astrophysical objects @xcite ( e.g. interiors of white dwarfs and magnetospheres of neutron stars and magnetars ) ; iv ) as well as for micro and nano - scale objects ( e.g. quantum diodes @xcite , quantum dots and nanowires @xcite , nano - photonics @xcite , plasmonics @xcite , ultra - small electronic devices @xcite , and metallic nanostructures @xcite ) ; v ) micro - plasmas @xcite , and quantum x - ray free - electron lasers @xcite . furthermore , it should be stressed that a fermi degenerate dense plasma may also arise when a pellet of hydrogen is compressed to many times the solid density in the fast ignition scenario for inertial confinement fusion @xcite . since there is an impressive developments in the field of short pulse petawatt laser technology , it is highly likely that such plasma conditions can be achieved by intense laser pulse compression using powerful x - ray pulses . here
ultrafast x - ray thomson scattering techniques can be used to measure the features of laser enhanced plasma lines , which will , in turn , give invaluable informations regarding the equation of state of shock compressed dense matters .
recently , spectrally resolved x - ray scattering measurements @xcite have been performed in dense plasmas allowing accurate measurements of the electron velocity distribution function , temperature , ionization state , and of plasmons in the warm dense matter regime @xcite . this novel technique promises to access the degenerate , the closely coupled , and the ideal plasma regime , making it possible to investigate extremely dense states of matter , such as the inertial confinement fusion fuel during compression , reaching super - solid densities . in this review article , we present the theoretical progress that has been recently made in the area of collective nonlinear interactions in collisionless dense quantum plasmas .
the manuscript is organized in the following fashion . in section 2 , we shall briefly recapitulate the hydrodynamic representation of some quantum - like models that appear in different branches of physics .
the governing equations for nonlinearly interacting electrostatic waves in an unmagnetized quantum plasma are derived in section 3 .
section 4 presents numerical studies of nonlinear electron plasma wave excitations in the form of quantized one - dimensional dark solitons and quantized two - dimensional vortices .
the model used here is the nonlinear schrdinger equation for the dispersive epos , coupled with the poisson equation for the electrostatic potential .
this model is also used for studying 3d quantum electron fluid turbulence in section [ sec : turbulence ] , where we find non - kolmogorov - type turbulence spectra . in section [ sec :
kinetic ] , we present recent results concerning the phase space ( kinetic ) turbulence , by using the wigner and vlasov models for the electron distribution function .
a theoretical model for the generation of quasi - stationary magnetic fields in a dense quantum plasma due to the weibel instability is presented in section [ sec : weibel ] .
the magnetization of a dense plasma in the presence of a large amplitude electromagnetic wave is demonstrated in subsection [ sec : magnetization ] .
the dynamics of electromagnetic waves in a dense magnetoplasma are discussed in section [ emwaveequation ] . here
we focus on spin waves propagating across the magnetic field direction , and develop nonlinear equations for low - phase speed ( in comparison with the speed of light ) electromagnetic waves in a dense quantum magnetoplasma .
finally , section [ sec : conclusions ] highlights our main results and describes the future prospectives of the quantum plasma physics research .
this section is included to show how different types of quantum - like models can be cast in the form of hydrodynamic equations .
first , we consider the non - stationary nonlinear schrdinger equation(nlse ) @xmath86 where @xmath87 is the macroscopic wave function , @xmath88 is the particle mass , and @xmath89 is an effective potential .
the nlse also arises in various physical context in the description of amplitude modulated nonlinear waves in fluids @xcite , in transmission lines @xcite , in nonlinear optics for ultra - fast communications @xcite , in plasmas @xcite , and in many other areas of physics @xcite .
introduce the madelung transformation @xcite @xmath90 where @xmath91 and @xmath92 are real , and obtain from ( 17 ) a pair of quantum hydrodynamic equations composed of the continuity and momentum equations , respectively , @xmath93 and @xmath94 .
\label{mom}\ ] ] here @xmath95 corresponds to the local density per unit length , and @xmath96 .
the quantum potential is @xmath97 we note that the quantum particle number density @xmath91 and the quantum velocity field @xmath98 can be written as , respectively , @xmath99 and @xmath100.\ ] ] where the asterisk denotes the complex conjugate .
the quantum - like velocity , given by ( 23 ) , is a potential field , namely , @xmath101 everywhere in a single - connected region .
let us now define the generalized vorticity on the weighted velocity field as @xcite @xmath102 where the first term in the right - hand side in ( 25 ) represents the ordinary vorticity .
it is well known @xcite that in the condensate state all rotational flow is carried by quantized vortices ( the circulation of the velocity around the core of each such vortex is quantized ) . in the absence of quantized vortices ,
the first term is zero in view of ( 24 ) .
in such a situation , the second term in ( 25 ) determines the generalized vorticity .
various aspects of quantized vortex dynamics and superfluid turbulence appear in barenghi et al @xcite .
bewley et al @xcite have presented a technique for visualization of quantized vortices in liquid helium .
second , the nonlinear schrdinger equation can be generalized by including the trapping potential @xmath103 which confines identical bosons @xcite in the harmonic trap of an ultracold quantum system . here
@xmath104 is the boson mass , and @xmath105 , @xmath106 and @xmath107 are the harmonic frequencies of the bosons along the @xmath108 and @xmath109 directions , respectively , of a cartesian coordinate system .
the nonlinear dynamics of the bose - einstein condensates ( becs ) @xcite is then governed by the gross - pitaevskii equation @xcite @xmath110 where for the becs we have @xmath111 . here
@xmath104 is the mass of the bosons and @xmath112 is the scattering length for boson - boson collisions .
the becs are repulsive ( attractive ) for @xmath113 @xmath114 . introducing the ansatz
@xmath115 $ ] in ( 27 ) , we obtain generalized quantum hydrodynamic equations @xcite @xmath116 and @xmath117 where the particle flux @xmath118 with @xmath119 .
equation ( 29 ) establishes the irrotational nature of the superfluid motion of the becs .
equations ( 29 ) and ( 30 ) can be used to study the linear and nonlinear properties of becs .
next , we consider the dynamics of a non - relativistic single fermi ( @xmath120 spin ) particle ( a degenerate electron ) governed by the pauli equation @xcite
@xmath121\psi = 0 , \label{schrod2}\ ] ] where @xmath122 is the wave function of the single particle species having the spin @xmath123 , @xmath124 is the pauli spin matrices , @xmath125 is the vector potential , @xmath126 is the scalar potential , @xmath127 , and @xmath128 is the bohr magneton . by using the madelung representation for the complex wave function @xcite @xmath129,\ ] ]
one can obtain from ( 31 ) the quantum magnetohydrodynamic equation @xcite @xmath130 and @xmath131 + \frac{\hbar^2}{2m_e } \nabla\left(\frac{\nabla^2 \sqrt{n_e}}{\sqrt{n_e}}\right ) + \mu_b \nabla ( { \boldsymbol \sigma } \cdot { \bf b}),\ ] ] where @xmath132 mimics the spinor through which the electron-@xmath133 spin properties are mediated , @xmath134 represents the probability density of finding the single electron at some point with a spin @xmath135 .
we have denoted the generalized electron momentum @xmath136 .
we can now express the quantum electron velocity @xcite @xmath137 the spin density vector @xmath138 and the spin vector transport equation @xcite @xmath139,\ ] ] where we have used the summation convention for repeated indices , and have denoted @xmath140 is the total derivative .
the electron momentum and electromagnetic fields are coupled via the maxwell equations . ignoring the electromagnetic fields and the particle spin , one can obtain , after linearizing ( 33 ) and ( 34 ) , the frequency of electron oscillations @xmath141 where @xmath142 is the wave number . finally , we consider interaction of an electron with both background electrons and singly charged positive ions .
the electron dynamics is governed by @xcite @xmath143 where @xmath144 is the potential responsible for the interaction of an electron with background matter which includes electrons and positively charged ions with the number density @xmath145 .
the wave function is normalized on the number density of electrons , viz .
we note that eq .
( 39 ) accounts only for the coulomb interactions between electrons and ions , and completely ignores the quantum statistical pressure , the self - consistent ambipolar field arising from the charge separation , and the electron spin-1/2 effect .
assume that @xmath147 , where @xmath148 represent the unperturbed electron number density and the perturbation wave function @xmath149 for spherically symmetric oscillations has the form @xmath150 , with @xmath151 being the normalization constant .
thus , the dispersion relation deduced from ( 39 ) reads @xcite @xmath152.\ ] ] the interaction between two electrons participating in spherically symmetric electron oscillations has been considered in ref .
the latter predicts that there would be an effective attraction between electrons mediated by low energy [ given by the minus sign in eq .
( 41 ) ] spherically symmetric oscillations of electrons in a quantum plasma .
we can thus have attracting degenerate electrons in dense plasmas .
the underlying physics of electron attraction here seems to be similar to that of the cooper pairing of electrons in superconductors in which electrons close to the fermi level attract each other due to their interactions with crystal lattice vibrations ( phonon oscillations ) .
the pairs of electrons act more like bosons which can condensate into the same energy level , contrary to single electrons which are fermions and must obey the pauli exclusion principle .
in the preceding section , we have seen that the quantum madelung fluid description predicts a diffraction pattern of a single electron or positron
. however , collective interactions between an ensemble of degenerate electrons ( fermions ) in dense plasmas are responsible for new linear and nonlinear waves and structures .
the quantum @xmath51-body problem is governed by the schrdinger equation for the @xmath51-particle wave function @xmath153 , where @xmath154 is the coordinate ( space , spin ) of particle @xmath155 . for identical fermions ,
the equilibrium @xmath51-particle wave function is given by the slater determinant @xcite @xmath156 which is anti - symmetric under odd numbers of permutations .
hence , @xmath157 vanishes if two rows are identical , which is an expression of the pauli exclusion principle that two identical fermions can not occupy the same state .
example @xmath158 : @xmath159 $ ] so that @xmath160 and @xmath161 . due to the pauli exclusion principle
, all electrons are not permitted to occupy the lowest energy state , and in the ultra - cold limit when all energy states up to the fermi energy level are occupied by electrons , there is still a quantum - statistical pressure determined by the fermi pressure . to describe collective electrostatic oscillations in a plasma ,
the quantum analogue of the vlasov - poisson system is the wigner - poisson system , given by @xmath162f({\bf x},{\bf v}',t)\ , d^3\lambda\ , d^3v ' , \label{eq43}\ ] ] and @xmath163 assuming immobile ions .
we note ( see appendix a ) that the wigner equation converges to the vlasov equation for classical particles ( electrons ) when @xmath164 @xmath165 we now take the moments of the wigner equation ( [ eq43 ] ) and obtain [ up to @xmath166 the non - relativistic quantum - electron fluid ( or the quantum madelung fluid ) equations @xcite composed of the electron continuity equation @xmath167 the electron momentum equation including the quantum statistical pressure and the quantum force @xmath168 where @xmath126 is determined from the poisson equation @xmath169 for the degenerate fermi - dirac distributed plasma , one has ( up to constants of order unity ) the quantum statistical pressure for the electrons @xmath170 where @xmath171 is the number of degrees of freedom in the system .
it should be noted that eliasson and shukla @xcite and tsintsadze and tsintsadze @xcite have obtained different expressions for @xmath172 in the non - zero limit of the electron fermi temperature .
the quantum force @xcite due to electron tunneling through the bohm potential is @xmath173 where @xmath174 represents the bohm potential .
we note that the @xmath175-term [ cf .
( 14 ) ] does not appear in ( 50 ) due to consideration of the term up to @xmath176 in the expansion parameter , as it also happens when the mean - field approximation @xcite is used in deducing @xmath177 . by introducing the wave function @xmath178 where
@xmath179 is defined according to @xmath180 and @xmath181 , it can be shown that the qhd equations [ e.g. eqs .
( 46)(48 ) ] are equivalent to the generalized nls - poisson system @xcite @xmath182 and @xmath183 the derivation of ( 52 ) required the electron plasma flow velocity to be curl free everywhere in a singly - connected region , except at points where the electron number density vanishes .
this is obviously not valid in general .
similar to quantum - fluid treatment , one should include the generalized electron vorticity @xmath184 , which is non - vanishing [ see the discussions below eq .
equation ( [ nls ] ) captures the two main properties of a quantum plasma , namely the quantum statistical pressure ( fully nonlinear ) and quantum dispersion effects , and is coupled self - consistently to the electrostatic potential via the poisson equation ( [ poisson ] ) .
we thus have a nonlocal nonlinear interaction between the electron density and the electrostatic potential .
furthermore , we note that one - dimensional version of eq .
( [ nls ] ) without the @xmath126-term has also been used to describe the behavior of a bose - einstein condensate @xcite in the absence of the confining potential .
linearization of the nls - poisson equations yields the frequency of the epos @xcite @xmath185 one can identify two distinct dispersive effects from ( 54 ) : one long wavelength regime , @xmath186 , and one short wavelength regime , @xmath187 separated by a critical wavenumber @xmath188 .
the dynamics of low - phase speed ( in comparison with the electron fermi speed ) nonlinear electrostatic ion oscillations in a quantum electron - ion plasma is governed by the inertialess electron equation of motion @xmath189 the ion continuity @xmath190 the ion momentum equation @xmath191 and the poisson equation @xmath192 where @xmath193 is the ion number density , @xmath194 is the ion fluid velocity , and @xmath195 is the ion charge state . in the quasi - neutral approximation ,
@xmath196 , we can combine eqs .
( [ electron ] ) and ( [ ionmomentum ] ) to obtain @xmath197 equations ( [ ioncontinuity ] ) and ( [ ionmomentum1 ] ) are the desired set for studying nonlinear ion waves @xcite in a dense quantum plasma .
we are now in a position to discuss nonlinear properties and the dynamics of localized electrostatic excitations in a dense quantum plasma based on the nonlinear equations we have developed in the preceding section . in this subsection
, we shall discuss the formation of one - dimensional quantized dark solitons and two - dimensional ( 2d ) quantized vortices associated with the epos in a dense quantum plasma at nanoscales ( of the order of @xmath198 ) .
we note that the dynamics of dark solitons in the 2d nonlinear schrdinger equation in a defocusing medium has been studied by ivonin @xcite and ivonin et al .
however , for studying the formation and dynamics of electrostatic nanostructures in a dense quantum plasma , we shall use the nonlinear schrdinger - poisson equations @xcite @xmath199 and @xmath200 where the normalized wave function @xmath201 is @xmath202 , the normalized @xmath203 is @xmath204 , and the time and space variables are in units of @xmath205 and @xmath198 , respectively .
we have denoted @xmath206 ( see ref .
@xcite ) .
the system of equations([eq1 ] ) and ( [ eq2 ] ) is supplemented by the maxwell equation @xmath207 where the electric field is @xmath208 .
the system ( [ eq1])([eq3 ] ) has the following conserved integrals : the number of electrons @xmath209 the electron momentum @xmath210 the electron angular momentum @xmath211 and the total energy @xmath212 \,d^3 x.\ ] ] the conserved quantities are required to check the accuracy of the numerical integration of ( 60 ) and ( 61 ) . for quasi - stationary one - dimensional structures moving with a constant speed @xmath213 , one can find localized , solitary wave solutions by introducing the ansatz @xmath214 , where @xmath215 is a complex - valued function of the argument @xmath216 , and @xmath217 and @xmath218 are a constant wavenumber and the frequency shift , respectively . by the choice @xmath219 , the coupled system of equations can be written as @xmath220 and @xmath221 where @xmath222 is an eigenvalue of the system . from the boundary conditions
@xmath223 and @xmath224 at @xmath225 , we determine @xmath226 and @xmath227 .
the system of eqs .
( [ eq4 ] ) and ( [ eq5 ] ) admits a first integral in the form @xmath228 where the boundary conditions @xmath223 and @xmath224 at @xmath225 have been employed .
( the upper panel ) and electrostatic potential @xmath203 ( the lower panel ) associated with a dark soliton supported by the system of equations ( [ eq4 ] ) and ( 68 ) , for @xmath229 ( solid lines ) , @xmath230 ( dashed lines ) , and @xmath231 ( dash - dotted line ) .
after ref .
@xcite.,width=302 ] ( left - hand panel ) and electrostatic potential @xmath203 ( the right - hand panel ) , obtained from a simulation of the system of equations ( [ eq1 ] ) and ( [ eq2 ] ) .
the initial condition is @xmath232\exp(i k x)$ ] , with @xmath219 , @xmath229 and @xmath233 .
after ref .
@xcite.,width=302 ] figure [ fig1 ] shows profiles of @xmath234 and @xmath203 obtained numerically from ( [ eq4 ] ) and ( [ eq5 ] ) for a few values of @xmath235 , where @xmath215 was set to @xmath236 on the left boundary and to @xmath237 on the right boundary , i.e. the phase shift is 180 degrees between the two boundaries .
the solutions are in the form of dark solitons , with a localized depletion of the electron density @xmath238 , associated with a localized positive potential .
larger values of the parameter quantum coupling parameter @xmath235 give rise to larger - amplitude and wider dark solitons .
the solitons localized
` shoulders' on both sides of the density depletion .
numerical solutions of the time - dependent system of eqs .
( [ eq1 ] ) and ( [ eq2 ] ) is displayed in fig .
[ fig2 ] , with initial conditions close ( but not equal ) to the ones in fig .
two very clear and long - lived dark solitons are visible , associated with a positive potential of @xmath239 , in agreement with the quasi - stationary solution of fig .
[ fig1 ] for @xmath229 .
in addition there are oscillations and wave turbulence in the time - dependent solution presented in fig .
hence , the dark solitons seem to be robust structures that can withstand perturbations and turbulence during a considerable time .
( upper panel ) and electrostatic potential @xmath203 ( lower panel ) associated with a two - dimensional vortex supported by the system ( [ eq7 ] ) and ( [ eq8 ] ) , for the charge states @xmath240 ( solid lines ) , @xmath241 ( dashed lines ) and @xmath242 ( dash - dotted lines ) .
we used @xmath229 in all cases .
after ref .
@xcite.,width=302 ] ( left panel ) and an arrow plot of the electron current @xmath243 ( right panel ) associated with singly charged ( @xmath244 ) two - dimensional vortices , obtained from a simulation of the time - dependent system of equations ( [ eq1 ] ) and ( [ eq2 ] ) , at times @xmath245 , @xmath246 , @xmath247 and @xmath248 ( upper to lower panels ) .
we used @xmath229 .
the singly charged vortices form pairs and keep their identities . after ref .
@xcite.,width=302 ] ( left panel ) and an arrow plot of the electron current @xmath243 ( right panel ) associated with double charged ( @xmath249 ) two - dimensional vortices , obtained from a simulation of the time - dependent system of eqs .
( [ eq1 ] ) and ( [ eq2 ] ) , at times @xmath245 , @xmath246 , @xmath247 and @xmath248 ( upper to lower panels ) .
we used @xmath229 .
the doubly charged vortices dissolve into nonlinear structures and wave turbulence .
after ref .
@xcite.,width=302 ] for the two - dimensional ( @xmath250 ) system , it is possible to find vortex structures of the form @xmath251 , where @xmath252 and @xmath253 are the polar coordinates defined via @xmath254 and @xmath255 , @xmath218 is a constant frequency shift , and @xmath256 for different excited states ( charge states ) . the index @xmath132 is also known as the circulation number @xcite . with this ansatz , eqs .
( [ eq1 ] ) and ( [ eq2 ] ) can be written in the form @xmath257\psi=0 , \label{eq7}\end{aligned}\ ] ] and @xmath258 respectively , where the boundary conditions @xmath259 and @xmath260 at @xmath261 determine the constant frequency @xmath262 .
different signs of charge state @xmath132 describe different rotation directions of the vortex . for @xmath263
, one must have @xmath264 at @xmath265 , and from symmetry considerations one has @xmath266 at @xmath265 .
figure [ fig3 ] shows numerical solutions of eqs .
( [ eq7 ] ) and ( [ eq8 ] ) for different @xmath132 and for @xmath229 . here
the vortex is characterized by a complete depletion of the electron density at the core of the vortex , and is associated with a positive electrostatic potential .
figure [ fig4 ] shows time - dependent solutions of eqs .
( [ eq1 ] ) and ( [ eq2 ] ) in two space dimensions for singly charged ( @xmath267 ) vortices , where , in the initial condition , four vortex - like structures were placed at some distance from each other .
the initial conditions were such that the vortices are organized in two vortex pairs , with @xmath268 , @xmath269 , @xmath270 , and @xmath271 , seen in the upper panels of fig .
the vortices in the pairs have opposite polarity on the electron fluid rotation , as seen in the in the upper right panel of fig .
interestingly , the `` partners '' in the vortex pairs attract each other and propagate together with a constant velocity , and in the collision and interaction of the vortex pairs ( see the second and third pairs of panels in fig . [ fig4 ] ) , the vortices keep their identities and change partners , resulting into two new vortex pairs which propagate obliquely to the original propagation direction . on the other hand , as shown in fig .
[ fig5 ] , vortices that are multiply charged ( @xmath272 ) are unstable . here
the system of eqs .
( [ eq1 ] ) and ( [ eq2 ] ) was again solved numerically with the same initial condition as the one in fig .
[ fig4 ] , but with doubly charged vortices @xmath273 , @xmath274 , @xmath275 , and @xmath276 .
the second row of panels in fig . [ fig5 ] reveals that the vortex pairs keep their identities for some time , while a quasi one - dimensional density cavity is formed between the two vortex pairs . at a later stage
, the four vortices dissolve into complicated nonlinear structures and wave turbulence .
hence , the nonlinear dynamics is very different between singly and multiply charged solitons , where only singly charged vortices are long - lived and keep their identities .
this is in line with previous results on the nonlinear schrdinger equation , where it was noted that vortices with higher charge states are unstable @xcite . in his classic paper , haas et al .
@xcite developed both small and large amplitude theories for one - dimensional solitary ion waves in a dense quantum plasmas .
they found that the dynamics of small amplitude solitary waves is governed by the kortweg de - vries ( k - dv ) equation @xmath277 where @xmath278 represents the relative ( with respect to @xmath279 ) ion density perturbation , the time and space variables are in units of the ion plasma period @xmath280 and the electron thomas - fermi radius @xmath281 , respectively .
the k - dv equation admits both the solitary and the periodic ( cnoidal ) waves @xcite . introducing the wave form @xmath282 ,
we can write ( 72 ) in the stationary frame as @xmath283 where @xmath284 represents the mach number , and @xmath285 is a constant of integration . in the special case when @xmath278 and its derivatives tend to zero at @xmath286 , @xmath287 .
multiplying ( 73 ) by @xmath288 one can integrate once the resultant equation , and express it in the form of an energy integral @xcite .
the resulting solitary wave solution of ( 73 ) is @xmath289 where @xmath290 and @xmath291 are the maximum amplitude and the width of the soliton .
we see that compressive solitary wave solutions are possible if @xmath292 .
there also exist possibility of one - dimensional large amplitude localized ion wave excitations . to demonstrate this
, we assume that the quantum force acting on the electrons dominates over the quantum statistical pressure , viz .
hence , the electron density is obtained from @xcite @xmath294 the electrons are coupled with ions through the space charge electric field ( @xmath295 ) .
the dynamics of singly charged ions is governed by the ion continuity @xmath296 and ion momentum equation @xmath297 where @xmath298 is @xmath299 component of the ion fluid velocity perturbation .
the system of eqs .
( [ qiaweq1])([qiaweq3 ] ) is closed by the poisson equation @xmath300 we now look for stationary nonlinear ion wave structures moving with a constant speed @xmath301 .
hence , all unknowns depend only on the variable @xmath302 .
defining @xmath303 , eq .
( [ qiaweq1 ] ) takes the form @xmath304 equations ( [ qiaweq2 ] ) and ( [ qiaweq3 ] ) can be integrated once with the boundary conditions @xmath305 and @xmath306 at @xmath307 , and the results can be combined to have @xmath308 inserting ( [ qiaweq6 ] ) into ( [ qiaweq4 ] ) we obtain @xmath309 equations ( [ qiaweq5 ] ) and ( [ qiaweq7 ] ) are the desired equations for studying the nonlinear ion waves in dense quantum plasmas .
it is convenient to introduce dimensionless quantities [ see below eq .
( [ qiaweq9 ] ) ] into eqs .
( [ qiaweq5 ] ) and ( [ qiaweq7 ] ) , and rewrite them as @xmath310 and @xmath311 where we have normalized the space variable as @xmath312 , the electron wave function as @xmath313 , and the potential as @xmath314 . here
@xmath315 is the quantum ion wave speed and @xmath316 is the quantum wavenumber .
the quantum `` mach number '' is defined as @xmath317 .
we note that the coupled eqs .
( [ qiaweq8 ] ) and ( [ qiaweq9 ] ) admit a conserved quantity @xmath318 where we have used the boundary conditions @xmath319 , and @xmath320 at @xmath321 .
for a symmetric solitary ion wave structure , we can assume that @xmath322 and @xmath323 , as well as @xmath324 at @xmath325 .
hence , at @xmath325 eq .
( [ qiaweq10 ] ) yields @xmath326 in the wave - breaking limit , where @xmath327 , we find that @xmath328 , or @xmath329 .
accordingly , the electron density will locally rise to twice the background density at wave breaking .
( top panel ) and the electron density @xmath330 ( bottom panel ) as a function of @xmath331 , for different values of the mach number : @xmath332 ( dashed curves ) , @xmath333 ( solid curve ) , and @xmath334 ( dotted curve ) . after ref .
@xcite.,width=321 ] in fig .
[ qiawfig1 ] , numerical solutions of eqs .
( [ qiaweq8 ] ) and ( [ qiaweq9 ] ) are displayed , showing profiles of the electrostatic potential and electron number densities for different values of @xmath335 .
we see that both the electrostatic potential and electron density have localized and strongly peaked maxima and an oscillatory tail .
the latter is in sharp contrast to the classical ( non - quantum ) case , where the ion acoustic solitary waves have a monotonic profile , which in the small amplitude limit , where the system is governed by the korteweg - de vries equation , assumes a secant hyperbolicus shape .
we observe from fig .
[ qiawfig1 ] that @xmath332 is close to the wave breaking limit above which there do not exist solitary wave solutions .
our numerical investigation also suggests that there is a lower limit of @xmath335 ( slightly lower than 0.75 ) , below which the solitary wave solution vanishes .
the statistical properties of turbulence and its associated electron transport at nanoscales in quantum plasmas has been investigated in both 2d and 3d by means of the coupled nls and poisson equations @xcite .
it has been found that the nonlinear coupling between the epos of different scale sizes gives rise to small - scale electron density structures , while the electrostatic potential cascades towards large - scales .
the total energy associated with the quantum electron plasma turbulence , nonetheless , processes a characteristic , _ non-_kolmogorov - like spectrum .
the electron diffusion caused by the electron fluid turbulence is consistent with the dynamical evolution of turbulent mode structures . to investigate the quantum electron fluid turbulence in 3d
, we use the nonlinear schrdinger - poisson equations @xcite @xmath336 and @xmath337 which govern the dynamics of nonlinearly interacting epos of different wavelengths . in eqs .
( [ eq1d ] ) and ( [ eq2d ] ) the wave function is normalized by @xmath338 , the electrostatic potential by @xmath339 , the time @xmath54 by the electron plasma period @xmath340 , and the space @xmath341 by the thomas - fermi debye radius @xmath198 .
we have introduced the notation @xmath342 .
the nonlinear mode coupling interaction studies are performed to investigate the multi - scale evolution of a decaying 3d electron fluid turbulence , which is described by eqs .
( [ eq1d ] ) and ( [ eq2d ] ) .
all the fluctuations are initialized isotropically ( no mean fields are assumed ) with random phases and amplitudes in fourier space , and evolved further by the integration of eqs .
( [ eq1d ] ) and ( [ eq2d ] ) , using a fully de - aliased pseudospectral numerical scheme @xcite based on the fourier spectral methods .
the spatial discretization in our 3d simulations uses a discrete fourier representation of turbulent fluctuations .
the numerical algorithm employed here conserves energy in terms of the dynamical fluid variables and not due to a separate energy equation written in a conservative form .
the evolution variables use periodic boundary conditions .
the initial isotropic turbulent spectrum was chosen close to @xmath343 , with random phases in all three directions .
the choice of such ( or even a flatter than @xmath344 ) spectrum treats the turbulent fluctuations on an equal footing and avoids any influence on the dynamical evolution that may be due to the initial spectral non - symmetry .
the equations are advanced in time using a second - order predictor - corrector scheme .
the code is made stable by a proper de - aliasing of spurious fourier modes , and by choosing a relatively small time step in the simulations .
our code is massively parallelized using message passing interface ( mpi ) libraries to facilitate higher resolution in a 3d computational box , with a resolution of @xmath345 grid points .
we study the properties of 3d fluid turbulence , composed of nonlinearly interacting epos , for two specific physical systems .
these are the dense plasmas in the next generation laser - based plasma compression ( lbpc ) schemes @xcite as well as in superdense astrophysical objects @xcite ( e.g. white dwarfs ) .
it is expected that in lbpc schemes , the electron number density may reach @xmath346 @xmath347 and beyond .
hence , we have @xmath348 s@xmath349 , @xmath350 erg , @xmath351 erg , and @xmath352 . the fermi debye length @xmath353 . on the other hand , in the interior of white dwarfs
, we typically have @xmath354 @xmath347 , yielding @xmath355 s@xmath349 , @xmath356 erg , @xmath357 erg , @xmath358 , and @xmath359 . the numerical solutions of eqs .
( [ eq1d ] ) and ( [ eq2d ] ) for @xmath360 and @xmath361 ( corresponding to @xmath362 @xmath347 and @xmath363 @xmath347 , respectively ) are displayed in figs .
[ fig1d ] and [ fig2d ] , respectively , which are the electron number density and electrostatic ( es ) potential distributions in the @xmath364-plane .
( top panels ) and @xmath365 ( bottom panels ) .
forward cascades are responsible for the generation of small - scale fluctuations seen in panels ( a ) and ( c ) .
large scale structures are present in the electrostatic potential , seen in panels ( b ) and ( d ) , essentially resulting from an inverse cascade . after ref .
@xcite.,width=302 ] is observed for @xmath366 .
the spectral index changes as a function of @xmath367 .
the numerical resolution is @xmath345 .
after ref .
@xcite.,width=302 ] figures [ fig1d ] and [ fig2d ] reveal that the electron density distribution has a tendency to generate smaller length - scale structures , while the es potential cascades towards larger scales .
the co - existence of the small and larger scale structures in turbulence is a ubiquitous feature of various 3d turbulence systems .
for example , in 3d hydrodynamic turbulence , the incompressible fluid admits two invariants , namely the energy and the mean squared vorticity .
the two invariants , under the action of an external forcing , cascade simultaneously in turbulence , thereby leading to a dual cascade phenomena . in these processes ,
the energy cascades towards longer length - scales , while the fluid vorticity transfers spectral power towards shorter length - scales . usually , a dual cascade is observed in a driven turbulence simulation , in which certain modes are excited externally through random turbulent forces in spectral space .
the randomly excited fourier modes transfer the spectral energy by conserving the constants of motion in @xmath142-space . on the other hand , in freely decaying turbulence
, the energy contained in the large - scale eddies is transferred to the smaller scales , leading to a statistically stationary inertial regime associated with the forward cascades of one of the invariants .
decaying turbulence often leads to the formation of coherent structures as turbulence relaxes , thus making the nonlinear interactions rather inefficient when they are saturated .
the power spectrum exhibits an interesting feature in our 3d electron plasma system , unlike the 3d hydrodynamic turbulence @xcite .
the spectral slope in the 3d quantum electron fluid turbulence is close to the iroshnikov - kraichnan power law @xcite @xmath368 , rather than the usual kolomogrov power law @xcite @xmath369 .
we further find that this scaling is not universal and is determined critically by the quantum tunneling effect .
for instance , for a higher value of h=1.0 the spectrum becomes more flat ( see fig [ fig2d ] ) .
physically , the flatness ( or deviation from the @xmath369 ) , results from the short wavelength part of the epos spectrum which is controlled by the quantum tunneling effect associated with the bohm potential .
the peak in the energy spectrum can be attributed to the higher turbulent power residing in the epo potential , which eventually leads to the generation of larger scale structures , as the total energy encompasses both the electrostatic potential and electron density components . in our dual cascade process
, there is a delicate competition between the epo dispersions caused by the statistical pressure law ( giving the @xmath370 term , which dominates at longer scales ) and the quantum bohm potential ( giving the @xmath371 term , which dominates at shorter scales with respect to a source ) .
furthermore , it is interesting to note that exponents other than @xmath369 have also been observed in numerical simulations @xcite of the charney and 3d incompressible navier - stokes equations .
recently , paoletti et al @xcite have examined the velocity statistics of quantum turbulence in superfluid @xmath372 , and found that it significantly differs from classical turbulence due to the topological interactions of vortices that are different from those in classical fluids . ,
@xmath373 and @xmath365 .
smaller values of @xmath367 corresponds to a small effective diffusion coefficient , which characterizes the presence of small - scale turbulent eddies that suppress the electron transport .
after ref .
@xcite.,width=302 ] we finally estimate the electron diffusion coefficient in the presence of small and large scale turbulent epos in our quantum plasma . an effective electron diffusion coefficient caused by the momentum transfer can be calculated from @xmath374 , where @xmath375 is electron momentum and the angular bracket denotes spatial averages and the ensemble averages are normalized to unit mass .
the effective electron diffusion coefficient , @xmath376 , essentially relates the diffusion processes associated with random translational motions of the electrons in nonlinear plasmonic fields .
we compute @xmath376 in our simulations , to measure the turbulent electron transport that is associated with the turbulent structures that we have reported herein .
it is observed that the effective electron diffusion is lower when the field perturbations are gaussian . on the other hand , the electron diffusion increases rapidly with the eventual formation of longer length - scale structures , as shown in fig .
[ fig3d ] .
the electron diffusion due to large scale potential distributions in quantum plasmas dominates substantially , as depicted by the solid - curve in fig .
[ fig3d ] .
furthermore , in the steady - state , nonlinearly coupled epos form stationary structures , and @xmath376 saturates eventually .
thus , remarkably an enhanced electron diffusion results primarily due to the emergence of large - scale potential structures in our 3d quantum plasma .
in the preceding sections , we have discussed the properties of quantized coherent structures and 3d quantum electron fluid turbulence based on the coupled schrdinger and poisson equations .
thus , it has been assumed that nonlinearly interacting plasma waves are spontaneously created by some known physical processes ( e.g. the beam - plasma instability ) in quantum plasmas .
the formation of electrostatic kinetic phase space structures , based on the vlasov - poisson equations , in classical plasmas has been well documented @xcite . in the following
, we shall discuss quasilinear aspects @xcite of the epos that are governed by the wigner - poisson system ( i.e. a quantum analogue of the vlasov - poisson system ) .
specifically , we focus on kinetic phase - space nonlinear structures arising from the trapping of electrons in the finite amplitude wave potential and the self - consistent modification of the electron distribution function in the presence of nano - kinetic structures . to study the differences in the nonlinear evolution of the wigner and vlasov equations , we have simulated the well - known bump - on - tail instability @xcite , whereby a high - velocity beam is used to destabilize a maxwellian equilibrium .
we use the initial condition @xmath377 $ ] , where @xmath378 represents random fluctuations of order @xmath379 that help seed the instability ( see fig .
@xmath380 is the electron thermal speed .
we use periodic boundary conditions with spatial period @xmath381 , where @xmath382 is the debye length .
three simulations were performed , with different values of the normalized planck constant , defined as @xmath383 : @xmath384 ( vlasov ) , @xmath360 , and @xmath385 .
, for a ) @xmath384 ( vlasov ) , b ) @xmath360 ( wigner ) , and c ) @xmath385 ( wigner ) .
initially monochromatic spectrum .
top panels : electron distribution function @xmath386 in phase space .
bottom panels : spatially averaged electron distribution function @xmath387 in velocity space . after ref .
@xcite.,width=510 ] in order to highlight the transient oscillations in velocity space , we first perturb the above equilibrium with a monochromatic wave having @xmath388 ( i.e. , a wavelength of @xmath389 ) .
figure [ fig2_quasi ] shows the results from simulations of the wigner - poisson and vlasov - poisson systems . in both simulations , due to the bump - on - tail instability ,
electrostatic waves develop nonlinearly and create periodic trapped - particle islands ( electron holes ) with the wavenumber @xmath390 .
the theory predicts the formation of velocity - space oscillations in the wigner evolution , which should be absent in the classical ( vlasov ) simulations .
this is the case in the results presented in fig .
2 , where the oscillations are clearly visible . , a ) for @xmath384 ( vlasov ) and b ) @xmath360 ( wigner ) . initially
broad wavenumber spectrum .
top panels : electron distribution function @xmath386 in phase space .
bottom panels : spatially averaged electron distribution function @xmath387 in velocity space . after ref .
@xcite.,width=340 ] when the initial excitation is broad - band ( i.e. , wavenumbers @xmath391 are excited ) , the electron holes start merging together at later times due to the sideband instability @xcite ( see fig . [ fig4_quasi ] ) . at this stage ,
mode coupling becomes important and quasilinear theory is not capable of describing these effects . as the system evolves toward larger spatial wavelength , the evolution becomes progressively more classical , with the appearance of a plateau in the resonant region .
nevertheless , at @xmath392 the wigner solution still displays some oscillatory behavior in velocity space , which is absent in the vlasov evolution . from the experimental viewpoint , recent collective x - ray scattering observations in warm dense matter
@xcite revealed a measurable shift in the plasmon frequency due to quantum effects . in the nonlinear regime ( strong excitations ) , this effect could lead to trapping of electrons in the wave potential of the plasmons , and the subsequent formation of the kind of phase space structures discussed here .
furthermore , we note that there also exists a theoretical description @xcite of quantum corrected electron holes based on the perturbative treatment of the wigner - poisson equations .
there are several mechanisms by which magnetic fields in classical plasmas can be generated .
they include i ) non - parallel density and temperature gradients ( the so called biermann battery @xcite ) , ii ) the electron temperature anisotropy ( known as the weibel instability @xcite ) , iii ) counterstreaming electron beams @xcite , and iv ) the ponderomotive forces of laser beams @xcite . in the following ,
we discuss two possibilities for the magnetic field generation in quantum plasmas .
we first discuss linear and nonlinear aspects @xcite of the weibel instability that is driven by equilibrium fermi - dirac electron temperature anisotropic distribution function in a nonrelativistic dense quantum plasma .
it is well known @xcite that a dense quantum plasma with an isotropic equilibrium distribution function does not admit any purely growing linear modes .
this can be verified , for instance , from the expression for the imaginary part of the transverse dielectric function , as derived by lindhard @xcite , for a fully degenerate non - relativistic fermi plasma .
it can be proven ( see eq .
( 30 ) of @xcite ) that the only exception would be for extremely small wavelengths , so that @xmath393 , where @xmath394 the characteristic fermi wave number of the system .
however , in this situation the wave would be super - luminal .
on the other hand , in a classical vlasov - maxwell plasma containing anisotropic electron distribution function , we have a purely growing weibel instability @xcite , via which dc magnetic fields are created .
the electron temperature anisotropy may arises due to the heating of the plasma by laser beams @xcite , where there is a signature of the weibel instability as well . in the next generation
intense laser - solid density plasma experiments , it is likely that electrons would be degenerate and that electron temperature anisotropy may develop due to an anisotropic electron heating by intense laser beams via resonant absorption , similar to the classical laser - plasma - interaction case @xcite . consider linear transverse waves in a dense quantum plasma composed of the electrons and immobile ions , with @xmath395 , where @xmath67 is the wave vector and @xmath396 is the wave electric field .
following the standard procedure , one then obtains the general dispersion relation @xcite for the transverse waves of the wigner - maxwell system @xmath397= 0 , \label{eq1weibel}\ ] ] where @xmath398 is the velocity vector , and @xmath399 is the equilibrium wigner function associated to fermi systems . for spin @xmath133 particles ,
the equilibrium pseudo distribution function is in the form of a fermi - dirac function .
here we allow for velocity anisotropy and express @xmath400 + 1 } , \label{eq2weibel}\ ] ] where @xmath16 is the chemical potential , and the normalization constant is @xmath401 here @xmath402 is a polylogarithm function @xcite . also , @xmath403 $ ] , where @xmath404 and @xmath405 are related to velocity dispersion in the direction perpendicular and parallel to @xmath109 axis , respectively . in the special case
when @xmath406 , the usual fermi - dirac equilibrium is recovered .
the chemical potential is obtained by solving the normalization condition ( [ eq3weibel ] ) , yielding , in particular , @xmath407 in the limit of zero temperature , where @xmath408 is the fermi energy .
also , the fermi - dirac distribution @xmath409 , where @xmath67 is the appropriated wave vector in momentum space , is related to the equilibrium wigner function ( [ eq2weibel ] ) by @xmath410 , with the factor @xmath411 coming from spin @xcite . however
, these previous works refer to the cases where there is no temperature anisotropy .
notice that it has been suggested @xcite that in laser plasmas the weibel instability is responsible for further increase of @xmath405 with time .
inserting ( [ eq2weibel ] ) into ( [ eq1weibel ] ) and integrating over the perpendicular velocity components , we obtain @xmath412 where @xmath413\bigr\ } - { \rm li}_{2}\bigl\{-\exp\bigl[-\bigl(\nu - \frac{h}{2}\bigl)^2 + \beta\mu\bigl]\bigr\}\biggr ) \ , .
\label{eq5weibel}\ ] ] in ( [ eq5weibel ] ) , @xmath414 is the dilogarithm function @xcite , @xmath415 is a characteristic parameter representing the quantum diffraction effect , @xmath416 , and @xmath417 , with @xmath418 . in the simultaneous limit of a small quantum diffraction effect ( @xmath419 ) and a dilute system ( @xmath420 )
, it can be shown that @xmath421 , where @xmath422 is the standard plasma dispersion function @xcite .
it is important to notice that either ( [ eq1weibel ] ) or ( [ eq4weibel ] ) reproduces the transverse dielectric function calculated from the random phase approximation for a fully degenerate quantum plasma @xcite , in the case of an isotropic system .
the simple way to verify this equivalence is to put @xmath423 in ( [ eq1weibel ] ) and then take the limit of zero temperature , so that @xmath424 for @xmath425 , and @xmath426 otherwise .
( @xmath427 ) and @xmath428 , relevant for the next generation inertially compressed material in intense laser - solid density plasma interaction experiments .
the temperature anisotropies are @xmath429 ( dashed line ) , @xmath430 ( solid line ) and @xmath431 ( dotted line ) , yielding , respectively , @xmath432 , @xmath433 and @xmath434 . after ref .
@xcite.,width=321 ] ( @xmath427 ) . here
the temperature anisotropy is @xmath430 .
we used @xmath435 ( dashed line ) , @xmath428 ( solid line ) and @xmath436 ( dotted line ) , yielding @xmath437 , @xmath438 and @xmath439 , respectively .
after ref .
@xcite.,width=321 ] we next solve our new dispersion relation ( [ eq4weibel ] ) for a set of parameters that are representative of the next generation laser - solid density plasma interaction experiments . the normalization condition ( [ eq3weibel ] ) can also be written as @xmath440=(4/3\sqrt{\pi})(\beta e_f)^{3/2}$ ] , which is formally the same relation holding for isotropic fermi - dirac equilibria @xcite . for a given value on the product @xmath441 and the density , this relation yields the value @xmath442 , from which the temperatures @xmath443 and @xmath444 can be calculated , if we know @xmath445 .
consider only purely growing modes . from the definition ( [ eq5weibel ] )
, one can show that @xmath446 when @xmath447 for a finite wavenumber @xmath142 . from ( [ eq4weibel ] )
we then obtain the maximum wavenumber for instability as @xmath448 . when @xmath449 , the range of unstable wavenumbers shrinks to zero .
in figs . [ fig1weibel ] and
[ fig2weibel ] , we have used the electron number density @xmath450 , which can be obtained in laser - driven compression schemes . the growth rate for different values on @xmath445
is displayed in fig .
[ fig1weibel ] .
we see that the maximum unstable wavenumber is @xmath448 , as predicted , and that the maximum growth rate occurs at @xmath451 .
figure [ fig1weibel ] also reveals that the maximum growth rate of the instability is almost linearly proportional to @xmath452 . in fig .
[ fig2weibel ] , we have varied the product @xmath441 , which is a measure of the degeneracy of the quantum plasma .
we see that for @xmath441 larger than @xmath453 , the instability reaches a limiting value , which is independent of the temperature , while thermal effects start to play an important role for @xmath441 of the order unity .
( top panel ) and @xmath454 ( bottom panel ) as a function of space and time , for @xmath428 and @xmath455 .
the magnetic field has been normalized by @xmath456 .
we see a nonlinear saturation of the magnetic field components at an amplitude of @xmath457 . after ref .
@xcite.,width=321 ] , over the simulation box ( top panel ) , and the logarithm of the magnetic field maximum ( bottom panel ) as a function of time , for @xmath430 and @xmath428 .
the magnetic field has been normalized by @xmath456 . from the logarithmic slope of the magnetic field in the linear regime one finds @xmath458 . after ref .
@xcite.,width=321 ] from several numerical solutions of the linear dispersion relation , it was deduced @xcite an approximate scaling law for the instability as @xmath459 , where the constant is approximately @xmath460 . using that @xmath461 , we have @xmath462 for the maximum growth rate of the weibel instability in a degenerate fermi plasma .
this scaling law , where the growth rate depends on the fermi energy and the temperature anisotropy , should be compared to that of a classical plasma @xcite , where the growth rate depends on the thermal energy and the temperature anisotropy . for a maxwellian plasma
, it has been found @xcite that the weibel instability saturates nonlinearly once the magnetic bounce frequency @xmath463 has increased to a value comparable to the linear growth rate . in order to assess the nonlinear behavior of the weibel instability for a degenerate plasma , we have carried out a kinetic simulation of the wigner - maxwell system .
we have assumed that the quantum diffraction effect is small , so that the simulation of the wigner equation can be approximated by simulations of the vlasov equation by means of an electromagnetic vlasov code @xcite . as an initial condition for the simulation
, we used the distribution function ( [ eq2weibel ] ) . in order to give a seed for any instability ,
the plasma density was perturbed with low - frequency fluctuations ( random numbers ) .
the results are displayed in figs .
[ fig3weibel ] and [ fig4weibel ] , for the parameters @xmath428 and @xmath430 , corresponding to the solid lines in figs .
[ fig1weibel ] and [ fig2weibel ] .
figure [ fig3weibel ] shows the magnetic field components as a function of space and time .
we see that the magnetic field initially grows , and saturates to steady state magnetic field fluctuations with an amplitude of @xmath464 .
the maximum amplitude of the magnetic field over the simulation box as a function of time is shown in fig .
[ fig4weibel ] , where we see that the magnetic field saturates at @xmath465 , while the linear growth rate of the most unstable mode is @xmath466 .
similar to the classical maxwellian plasma case @xcite , we can thus estimate the generated magnetic field as @xmath467 for a degenerate fermi plasma . for our parameters parameters relevant for intense laser - solid interaction experiments , we will thus have magnetic fields of the order @xmath468 ( one gigagauss ) .
the second example of the dense plasma magnetization can occur in the presence of streaming electrons and large amplitude electromagnetic waves . here
, the nonstationary ponderomotive force of the electromagnetic wave would create slowly varying electric fields and currents , which generate d.c . magnetic fields in dense plasmas .
let us consider the propagation of the electromagnetic wave with the electric field @xmath469 + c.c .
, in an unmagnetized non - relativistic dense plasma with streaming electrons ( with the drift velocity @xmath470 , where @xmath471 is the magnitude of the electron drift speed and @xmath472 is the unit vector along the @xmath109 axis in a cartesian coordinate system ; typically @xmath471 is much smaller than the electron fermi speed ) and immobile ions . here , @xmath473 is the envelope of the electromagnetic field at the position @xmath341 and time @xmath54 , and @xmath474 stands for complex conjugate .
the frequency @xmath65 and the wave vector @xmath475 , where @xmath476 is the unit vector along the @xmath299 axis , are related by @xcite @xmath477 where @xmath51 is the index of refraction and @xmath478 , the electromagnetic wave exerts a ponderomotive force @xmath479 on the plasma electrons , where the stationary and non - stationary ponderomotive forces @xcite are , respectively , @xmath480 and @xmath481}{\partial \omega } \frac{\partial |{\bf e}_0|^ 2}{\partial t}. \label{eq_m3}\ ] ] the ponderomotive force pushes the electrons locally , and creates the slowly varying electric field @xmath482 where the scalar and vector potentials are , respectively , @xmath483 and @xmath484}{\partial \omega } |{\bf e}_0|^ 2 .
\label{eq_m6}\ ] ] the induced slowly varying magnetic field @xmath485 is then @xmath486 . noting that @xmath487}{\partial \omega } = \frac{2 \omega k^ 2 u^ 2 \omega_{pe}^ 2 } { ( \omega^ 2- k^ 2 v_{fe}^ 2 -\omega_q^ 2)^ 2 } , \label{eq_m7}\ ] ] we can express the magnitude of the magnetic field as @xmath488 where @xmath489 is scale length of the envelope @xmath490 and @xmath491^{1/2 } $ ] .
we note from ( [ eq_m8 ] ) that the magnetic field strength is proportional to @xmath492 , and attains a large value when @xmath493 .
the electron gyrofrequency @xmath494 is @xmath495 where @xmath496 is the electron quiver velocity in the electromagnetic field .
we here consider various electromagnetic ( em ) wave modes and their nonlinear interaction in dense plasmas . as examples
, we will discuss spin waves , which occur in a magnetized plasma due to the spin-1/2 effect of electrons and positrons .
we also consider nonlinear interactions between finite amplitude electromagnetic and electrostatic waves in dense plasmas .
we focus on the underlying physics of stimulated scattering instabilities and the wave localization due to the parametric interactions involving the radiation pressure @xcite . in the presence of an external magnetic field
, the quantum description of the linear kinetics of a dense collisionless plasma is complicated due to quantization of the gyromagnetic motion and the inclusion of the electron and positron spin-@xmath133 . in the past
, many authors @xcite investigated the high - frequency conductivity , longitudinal and transverse dielectric responses in dense magnetized plasmas .
the propagation characteristics of high - frequency electromagnetic waves in quantum magnetoplasmas is different from that in a classical magnetoactive plasma @xcite .
recently , oraevsky et al .
@xcite found a new electromagnetic spin wave whose electric field is parallel to @xmath497 , and which propagates across @xmath472 , where @xmath472 is the unit vector along the @xmath109 axis in a cartesian coordinate system and @xmath498 is the strength of the ambient magnetic field .
the spin wave accompany the magnetization current due to spinning electron motion and the wave frequency is obtained from @xcite @xmath499 where @xmath500 is the perpendicular component of the wave vector @xmath501 , @xmath476 is the unit vector transverse to @xmath472 , @xmath502 is the electron collision frequency , nd @xmath503 . the third term in the right - hand side of ( 102
) represents the electron spin - magnetic resonance at the electron gyrofrequency @xmath504 . assuming that @xmath505
, we obtain from ( 102 ) @xmath506.\ ] ] for @xmath507 , the damping rate of the spin mode is @xmath508 the ponderomotive force of the spin wave can create compressional magnetic field perturbation due to an inverse cotton - mouton / faraday effect @xcite
. finite amplitude electromagnetic waves in quantum magnetoplasmas interact nonlinearly among themselves . in this subsection , we shall use the generalized q - mhd equations to obtain compact nonlinear equations for the electron magnetohydrodynamic ( emhd ) and the hall - mhd plasmas , and show how does the density , the fluid velocity , and magnetic field perturbations are coupled in a non - trivial manner .
the governing nonlinear equations for the electromagnetic waves in dense magnetoplasmas are the quantum magnetohydrodynamic equations composed of the continuity equation @xmath509 the electron and ion momentum equations , respectively , @xmath510 @xmath511 the faraday law @xmath512 the maxwell equation including the magnetization spin current @xmath513 where the pressure for non - relativistic degenerate electrons reads @xcite @xmath514 , \label{eq_em6}\ ] ] where @xmath515 , and the value of @xmath516 is fixed by the largest integer @xmath517 that satisfies @xmath518 .
the sum of the quantum bohm and intrinsic angular momentum spin forces is @xmath519 in eqs .
( [ eq_em1])([eq_em7 ] ) , @xmath520 is the number density of the particle species @xmath155 ( @xmath155 equals @xmath21 for the electrons , and @xmath521 for the ions ) , @xmath522 is the particle fluid velocity , and @xmath523 .
we have denoted the plasma current density @xmath524 and the electron magnetization spin current density @xmath525 , where the magnetization for dynamics on a time scale much slower than the spin precession frequency for an electron fermi gas reads @xcite @xmath526 .
first , we present the generalized nonlinear emhd equations for a dense magnetoplasma . here
the ions form the neutralizing background .
the wave phenomena in the emhd plasma will occur on a time scale much shorter than the ion plasma and ion gyroperiods . in equilibrium , we have @xmath527 the relevant nonlinear emhd equations are @xmath528 the electron momentum equation ( [ eq_em2 ] ) , faraday s law ( [ eq_em3 ] ) , and the electron fluid velocity given by @xmath529 we observe that the quantum tunneling and spin forces play an important role if there are slight electron density and magnetic field inhomogeneities in dense plasmas .
the nonlinear emhd equations are useful for studying collective electron dynamics in metallic and semiconductor nanostructures @xcite .
second , we derive the modified nonlinear hall - mhd equations in a dense electron - ion plasma .
the hall - mhd equations shall deal with the wave phenomena on a time scale larger than the electron gyroperiod .
the relevant nonlinear hall - mhd equations are the electron and ion continuity equations , the inertialess electron momentum equation @xmath530 faraday s law ( [ eq_em3 ] ) , the ion momentum equation @xmath531 and the electron fluid velocity given by @xmath532 where we have neglected the displacement current , since the hall - mhd plasma deals with electromagnetic waves whose phase velocity is much smaller than the speed of light in vacuum .
we now eliminate the electric field from ( [ eq_em13 ] ) by using ( [ eq_em12 ] ) , obtaining @xmath533 . \label{eq_em16}\ ] ] furthermore , eliminating @xmath534 from ( [ eq_em16 ] ) by using ( [ eq_em15 ] ) , we have @xmath535 \times { \bf b } - \frac{\nabla p_e}{n_e e } + \frac{{\bf f}_{qe}}{n_e e } \right\ } , \label{eq_em17}\ ] ] with the quasi - neutrality condition @xmath536 . finally , by using ( [ eq_em12 ] ) we can eliminate @xmath396 from ( [ eq_em2 ] ) , obtaining @xmath537 . \label{eq_em18}\ ] ] the ion continuity equation , eqs .
( [ eq_em17 ] ) and ( [ eq_em18 ] ) , together with ( [ eq_em13 ] ) and @xmath538 , where @xmath539 are the desired generalized nonlinear equations for the low - frequency ( in comparison with the electron gyrofrequency ) , low - phase velocity ( in comparison with the speed of light in vacuum ) in a hall - mhd dense plasma .
they can be used to investigate the multi - dimensional linear and nonlinear waves ( e.g. magnetosonic solitons @xcite ) , as well as nanostructures and turbulences @xcite in dense quantum magnetoplasmas .
nonlinear interactions between the high - frequency em waves and low - frequency electrostatic waves give rise to stimulated scattering instabilities in classical plasmas @xcite .
there also exists possibility of exciting plasma waves by the high - frequency em waves in quantum plasmas due to the parametric instabilities @xcite .
the governing equations for the high - frequency electromagnetic waves @xcite and the radiation pressure driven modified langmuir and ion - acoustic oscillations in an unmagnetized quantum plasma are , respectively , @xmath540 @xmath541 and @xmath542 where @xmath125 is the vector potential of the high - frequency electromagnetic wave , and @xmath543 is the electron density perturbation of the low - frequency oscillations ( viz .
the modified epos and ion waves ) . combining eqs .
( [ qem9])-([qem11 ] ) we thus simply obtain the nonlinear dispersion relations @xmath544 and @xmath545 where @xmath546 and @xmath547
. equations ( 126 ) and ( 127 ) admit stimulated raman , stimulated brillouin and modulational instabilities of the electromagnetic pump ( with the amplitude @xmath548 and the frequency @xmath549 ) in dense quantum plasmas .
we have denoted @xmath550 , where @xmath551 is the group velocity of the em pump wave frequency , and @xmath552 is the small frequency shift arising from the nonlinear interaction of the pump with the electrostatic perturbations @xmath553 in our quantum plasma . for stimulated raman and brillouin scatterings , the em pump wave is scattered off the resonant electron plasma wave and resonant ion wave , respectively , while for the modulational instability the electron and ion plasma oscillations are off resonant , so are the em sidebands . for three - wave decay interactions ,
we suppose that @xmath554 and @xmath555 .
thus , we ignore @xmath556 from ( 126 ) and ( 127 ) . letting @xmath557 and @xmath558 in ( 126 ) and ( 127 ) , respectively , and assuming @xmath559 , we obtain the the growth rates for stimulated raman and brillouin scattering instabilities ( denoted by the subscripts @xmath560 and @xmath561 , respectively , @xmath562 and @xmath563 where @xmath564 .
we note that the growth rates for stimulated raman and brillouin scattering instabilities are inversely proportional to the square roots of @xmath565 and @xmath566 , which depend on the quantum parameters . stimulated raman and brillouin scattering instabilities of the em wave off the es waves should provide invaluable information regarding the density fluctuations and equation of states that might exist in dense quantum plasmas .
the quantum corrected 3d zakharov equations @xcite have been derived by haas @xcite and haas and shukla @xcite , who demonstrated that the dispersive effects associated with quantum corrections can prevent the collapse of localized langmuir envelope electric fields in both two and three spatial dimensions .
nonlinear interactions between large amplitude electromagnetic waves and electrostatic plasma waves can produce nonlinear nanostructures composed of a density cavity that traps the electromagnetic wave envelope .
here we demonstrate trapping of intense electromagnetic waves into a finite amplitude density hole @xcite arising at the scale - size of the order of the electron skin depth @xmath567 ) .
a powerful circularly polarized electromagnetic ( cpem ) plane wave interacting nonlinearly with the epos would generate an envelope of the cpem vector potential @xmath568 , which shall obey the nonlinear schrdinger equation @xcite @xmath569 where the electron wave function @xmath157 and the scalar potential are governed by , respectively , @xmath570 and @xmath571 where the electron number density is defined as is given by the @xmath572 term . here
@xmath573 represents the cpem wave frequency , @xmath574 is the @xmath299 component of the group velocity of the cpem wave , @xmath575 is a quantum coupling parameter , and @xmath576 is the relativistic gamma factor due to the electron quiver velocity in the cpem wave fields .
the details of normalization of variables is given in ref .
the nonlinear coupling between intense cpem waves and epos comes about due to the nonlinear current density , which is represented by the term @xmath577 in eq .
( [ eq9 ] ) . in eq .
( [ eq10 ] ) , @xmath578 is the relativistic ponderomotive potential @xcite , which arises due to the cross - coupling between the cpem wave - induced electron quiver velocity and the cpem wave magnetic field .
the effect of quantum dispersion on localized electromagnetic pulses can be studied by considering a steady state structure moving with a constant speed @xmath574 . inserting the ansatz @xmath579 , @xmath580 and @xmath581 into eqs .
( [ eq9])([eq11 ] ) , where @xmath582 , @xmath583 and @xmath584 , and where @xmath585 and @xmath586 are real , one obtains from ( [ eq9])-([eq11 ] ) the coupled system of equations @xmath587 @xmath588 where @xmath589 , and @xmath590 with the boundary conditions @xmath591 and @xmath592 at @xmath225 . in eq .
( [ eq13 ] ) , @xmath593 represents a nonlinear frequency shift of the cpem wave . in the limit @xmath594 , one has from ( [ eq14 ] ) @xmath595 , where @xmath596 , and one recovers the classical ( non - quantum ) case of the relativistic solitary waves in a cold plasma @xcite .
the system of equations ( [ eq13])([eq15 ] ) admits a hamiltonian @xmath597 where the boundary conditions @xmath598 , @xmath599 and @xmath600 at @xmath225 have been used .
numerical solutions of the quasi - stationary system ( [ eq13])([eq15 ] ) are presented figs .
[ fig6 ] and [ fig7 ] , while time - dependent solutions of eqs .
( [ eq9])([eq11 ] ) are shown in figs .
[ fig8 ] and [ fig9 ] . here
parameters were used that are representative of the next generation laser - based plasma compression ( lbpc ) schemes @xcite .
the formula @xcite @xmath601 will determine the normalized vector potential , provided that the cpem wavelength @xmath602 ( in microns ) and intensity @xmath603 ( in w/@xmath604 ) are known .
it is expected that in lbpc schemes , the electron number density @xmath279 may reach @xmath346 @xmath347 and beyond , and the peak values of @xmath605 may be in the range 1 - 2 ( e.g. for focused em pulses with @xmath606 nm and @xmath607 w/@xmath608 ) . for @xmath348 s@xmath349 , one has @xmath609 erg and @xmath610 , since @xmath611 erg . the electron skin depth @xmath612 . on the other hand ,
a higher value of @xmath613 is achieved for @xmath614 s@xmath349 .
thus , our numerical solutions below , based on these two values of @xmath575 , have focused on scenarios that are relevant for the next generation intense laser - solid density plasma interaction experiments @xcite .
figures [ fig6 ] and [ fig7 ] show numerical solutions of eqs .
( [ eq13])([eq15 ] ) for several values of @xmath575 .
the nonlinear boundary value problem was solved with the boundary conditions @xmath599 and @xmath615 at the boundaries at @xmath616 .
one can see that the solitary envelope pulse is composed of a single maximum of the localized vector potential @xmath215 and a local depletion of the electron density @xmath617 , and a localized positive potential @xmath126 at the center of the solitary pulse .
the latter has a continuous spectrum in @xmath618 , where larger values of negative @xmath618 are associated with larger amplitude solitary em pulses . at the center of the solitary em pulse ,
the electron density is partially depleted , as in panels a ) of fig .
[ fig6 ] , and for larger amplitudes of the em waves one has a stronger depletion of the electron density , as shown in panels b ) and c ) of fig .
[ fig6 ] . for cases where the electron density goes to almost zero in the classical case @xcite ,
one important quantum effect is that the electrons can tunnel into the depleted region .
this is seen in fig .
[ fig7 ] , where the electron density remains nonzero for the larger value of @xmath575 in panels a ) , while the density shrinks to zero for the smaller value of @xmath575 in panel b ) .
( top row ) , the electron number density @xmath617 ( middle row ) and the scalar potential @xmath619 ( bottom row ) for @xmath620 ( left column ) , @xmath621 ( middle column ) and @xmath622 ( right column ) , with @xmath623 .
after ref .
@xcite.,width=302 ] ( top row ) , the electron number density @xmath617 ( middle row ) and the scalar potential @xmath619 ( bottom row ) for @xmath613 ( left column ) and @xmath623 ( right column ) , with @xmath621 .
after ref .
@xcite.,width=302 ] and the electron number density @xmath572 ( upper panels ) and of the electrostatic potential @xmath619 ( lower panel ) for @xmath623 . after ref .
@xcite.,width=302 ] and the electron number density @xmath572 ( upper panels ) and the electrostatic potential @xmath126 ( lower panel ) for @xmath613 . after ref .
, width=302 ] figures [ fig8 ] and [ fig9 ] show numerical simulation results of eqs .
( [ eq9])([eq11 ] ) , in order to investigate the quantum diffraction effects on the dynamics of localized cpem wavepackets . here the long - wavelength limit @xmath624 and @xmath625 was considered . in the initial conditions ,
an em pump with a constant amplitude @xmath626 and a uniform plasma density @xmath259 was used , together with a small amplitude noise ( random numbers ) of order @xmath627 added to @xmath628 to give a seeding any instability .
the numerical results are displayed in figs .
[ fig8 ] and [ fig9 ] for @xmath623 and @xmath613 , respectively . in both cases
, one can see an initial linear growth phase and a wave collapse at @xmath629 , in which almost all the cpem wave energy is contracted into a few well separated localized cpem wave pipes .
these are characterized by a large bell - shaped amplitude of the cpem wave , an almost complete depletion of the electron number density at the center of the cpem wavepacket , and a large - amplitude positive electrostatic potential . comparing fig .
[ fig8 ] with fig .
[ fig9 ] , one can see that there is a more complex dynamics in the interaction between the cpem wavepackets for the larger @xmath613 , shown in fig .
[ fig9 ] , in comparison with @xmath623 , shown in fig .
[ fig8 ] , where the wavepackets are almost stationary when they are fully developed .
in this paper , we have presented our up - to - date theoretical knowledge of nonlinear physics of non - relativistic quantum plasmas .
we started with nonlinear quantum models that describe the physics of localized excitations in different areas of physics .
we then moved to discussing the well known electron and ion plasma wave spectra in dense quantum plasmas , and presented nonlinear models for treating nonlinear interactions among finite amplitude plasma waves at nanoscales .
as examples , we demonstrated the existence of localized nonlinear epos and localized ion waves in dense quantum plasmas .
for electrostatic epos , the electron dynamics is governed by a pair composed of nonlinear schrdinger and poisson ( nlsp ) equations .
we stress that the nonlinear schrdinger equation governing the spatio - temporal evolution of the electron wave function in the presence of self - consistent electrostatic potential in dense plasmas has been obtained from the quantum electron momentum equation by introducing the eikonal representation as a mathematical tool for understanding the complex electron plasma wave interactions at nanoscales .
such a description also appears in the context of the nonlinear electron dynamics in thin metal films @xcite .
our nlsp equations admit a set of conserved quantities , e.g. the total number of electrons , the electron momentum , the electron angular momentum , and the electron energy . we have found that the nlsp equations admit quasi - stationary , localized structures in the form of one - dimensional quantized dark solitons and two - dimensional quantized vortices .
these nanostructures are associated with a local depletion of the electron density associated with positive electrostatic potential , and are parameterised by the quantum coupling parameter only . in the two - dimensional geometry ,
there exist a class of vortices of different excited states ( charge states ) associated with a complete depletion of the electron density and an associated positive potential .
numerical simulations of the time - dependent nlsp equations demonstrate the stability of stable dark solitons in one space dimension with an amplitude consistent with the one found from the time - independent solutions . in two - space - dimensions ,
the dark solitons of the first excited state were found to be stable and the preferred nonlinear state was in the form of vortex pairs of different polarities .
one - dimensional dark solitons and singly charge two - dimensional vortices are thus long - lived nonlinear structures in quantum plasmas .
similar to the bernstein - green - kruskal modes @xcite , we accounted for the trapping of electrons in the electrostatic wave potential and studied numerically the deformation of the equilibrium fermi - dirac distribution function and the subsequent emergence of localized phase space kinetic structures .
such an investigation was carried out by using the time - dependent wigner and poisson equations .
furthermore , we presented two mechanisms for the generation of magnetic fields in quantum plasmas .
they are associated with the quantum weibel instability in the presence of equilibrium anisotropic fermi - dirac electron distribution function , as well as with the the non - stationary ponderomotive force of a large amplitude electromagnetic waves in quantum plasmas with streaming electrons .
spontaneously generated magnetic fields can affect the linear and nonlinear propagation of both the electrostatic and electromagnetic waves in quantum magnetoplasmas @xcite .
the quantum corrections produce dispersion at short scales for the electrostatic upper - hybrid , lower - hybrid , and ion - cyclotron waves , while the quantum bohm force and the electron@xmath630 spin effects introduce new features to the elliptically polarized extraordinary electromagnetic mode @xcite .
furthermore , the electron-@xmath631 spin is responsible for a new perpendicularly propagating ( with respect to the magnetic field direction ) electromagnetic spin wave , which can be excited by intense neutrino bursts in supernovae .
the newly derived nonlinear emhd and hall - mhd equations can be used for investigating the multi - dimensional linear and nonlinear electromagnetic waves in quantum plasmas .
finally , we have also presented theoretical and numerical studies of stimulated raman and brillouin scattering instabilities of a large amplitude electromagnetic wave , and the trapping of arbitrary large amplitude circularly polarized em waves in a fully nonlinear electron density hole in an unmagnetized quantum plasma .
it is expected that localized nanostructures can transport electromagnetic wave energy over nanoscales in laboratory and astrophysical dense plasmas with degenerate electrons . the field of the nonlinear quantum plasma physics is extremely rich and vibrant today , and it holds a great promise of providing new practical technologies .
for example , the plasma assisted carbon nanostructures and nanomaterials are the future of nanotechnologies , so are the new radiation sources in the x - rays and gamma - rays regimes . in such circumstances
, one should have a complete understanding of the fundamentals of collective nonlinear interactions ( e.g. intense high - order harmonic generation of ultrashort laser pulses from laser - irradiated dense plasma surfaces ) in quantum plasmas .
furthermore , in magnetars , and in the next generation intense laser - solid density plasma interaction experiments , one would certainly have degenerate positrons , besides degenerate electrons .
the physics of dense quantum magnetoplasmas with degenerate electron - positron pairs is expected to be quite different than what has been described in this paper .
the reason is the complex nonlinear dynamics of the electron - positron pairs , which would have relativistic velocities in a dense magnetoplasma .
accordingly , we should develop new theories involving relativistic kinetic and quantum relativistic magnetohydrodynamic equations that include quantum relativistic effects @xcite , electromagnetic forces , angular momentum spin and nonlinear effects on equal footings .
a detailed analysis of such theories would provide us a guide line for understanding the origin of localized high - energy radiation and other complex phenomena ( e.g. the formation of structures ) from astrophysical settings and future laboratory experiments aiming to model astrophysical scenarios .
* acknowledgements * one of the authors ( p k shukla ) acknowledges the benefit of useful discussions with academician professor vladimir fortov .
we greately appreciate our invaluable collaborations with lennart stenflo , gert brodin , mattias marklund , fernando haas and nitin shukla .
this research was partially supported by the deutsche forschungsgemeinschaft ( bonn , germany ) through the project sh21/3 - 1 of the research unit 1048 , and by the swedish research council ( vr ) .
we here show that the wigner equation converges to the vlasov equation in the classical limit @xmath164 .
the wigner equation reads @xmath632 \left [ \phi\left({\bf x}+\frac{\boldsymbol{\lambda}}{2},t\right ) -\phi\left({\bf x}-\frac{\boldsymbol{\lambda}}{2},t\right)\right]f({\bf x},{\bf v}',t ) .
\label{w_eq1}\ ] ] by the change of variables @xmath633 we have @xmath634 and eq .
( [ w_eq1 ] ) takes the form @xmath635
\left [ \phi\left({\bf x}+\frac{\hbar\boldsymbol{\xi}}{2m_e},t\right ) -\phi\left({\bf x}-\frac{\hbar\boldsymbol{\xi}}{2m_e},t\right)\right]f({\bf x},{\bf v}',t ) .
\label{w_eq2}\ ] ] assuming that @xmath636 is small and taylor expanding @xmath126 up to the third order around @xmath637 , @xmath638 we have @xmath639 \left[\boldsymbol{\xi}\cdot\nabla\phi({\rm x},t)+\frac{\hbar^2}{24 m_e^2 } ( \boldsymbol{\xi}\cdot\nabla)^3\phi({\rm x},t)\right ] f({\bf x},{\bf v}',t ) .
\label{w_eq4}\ ] ] by the identity @xmath640\boldsymbol{\xi}=
i\nabla_{{\bf v}'}\exp[i ( { \bf v}-{\bf v}')\cdot{\boldsymbol{\xi}}],\ ] ] where @xmath641 , we have @xmath642\bigg [ ( \overleftarrow{\nabla}_{{\bf v}'}\cdot\overrightarrow{\nabla } ) -\frac{\hbar^2}{24 m_e^2}(\overleftarrow{\nabla}_{{\bf v}'}\cdot\overrightarrow{\nabla})^3\bigg ] \phi({\rm x},t)\bigg\ } f({\bf x},{\bf v}',t ) , \label{w_eq5}\ ] ] where the arrows indicate the direction of operation of the nabla operators . integration by parts in @xmath643 space , and using that @xmath644 for @xmath645 , now yields @xmath646\bigg\ { \phi({\bf x},t ) \bigg[(\overleftarrow{\nabla}\cdot\overrightarrow{\nabla}_{{\bf v } ' } ) -\frac{\hbar^2}{24 m_e^2}(\overleftarrow{\nabla}\cdot\overrightarrow{\nabla}_{{\bf v}'})^3\bigg ] f({\bf x},{\bf v}',t)\bigg\}. \label{w_eq6}\ ] ] now the integration in @xmath647 space can formally be performed , with the result @xmath648 f({\bf x},{\bf v}',t)\bigg\}. \label{w_eq7}\ ] ] where the identity @xmath649=(2\pi)^3\delta({\bf v}-{\bf v } ' ) \label{eq8}\ ] ] was used and where @xmath378 is dirac s delta function . finally ,
integration over @xmath643 space yields @xmath650 f({\bf x},{\bf v},t)\bigg\}. \end{split } \label{w_eq9}\ ] ] in the limit @xmath164 , we recover the vlasov equation @xmath651
we here present a derivation of the dispersion relation for electrostatic waves in a degenerate quantum plasma , governed by the wigner - poisson system of equations .
the linearized wigner - poisson system of equations reads @xmath652 \left [ \phi_1\left({\bf x}+\frac{\boldsymbol{\lambda}}{2},t\right ) -\phi_1\left({\bf x}-\frac{\boldsymbol{\lambda}}{2},t\right)\right]f_0({\bf v } ' ) , \label{eq_b1}\ ] ] and @xmath653 where @xmath654 denotes the background distribution function , and @xmath655 and @xmath656 denote the perturbed distribution function and electrostatic potential , respectively .
fourier transforming ( [ eq_b1])-([eq_b2 ] ) in space and laplace transforming them in time , we obtain @xmath657 \left [ e^{i { \bf k}\cdot\boldsymbol{\lambda}/2}-e^{-i { \bf k}\cdot\boldsymbol{\lambda}/2 } \right]f_0({\bf v}')\phi_1(\omega,{\bf k } ) , \label{eq_b3}\ ] ] @xmath658 rewriting ( [ eq_b3 ] ) as @xmath659 \right . \\ & \left
- \exp\left[i \frac{m_e}{\hbar } ( { \bf v}-{\bf v } ' ) \cdot{\boldsymbol{\lambda}}-i { \bf k}\cdot\boldsymbol{\lambda}/2\right ] \right\}f_0({\bf v}')\phi_1(\omega,{\bf k } ) , \end{split}\ ] ] and performing the integration over @xmath660 space , we have @xmath661 \right . \left .
-\delta\left[\frac{m_e}{\hbar } ( { \bf v}-{\bf v}')-\frac{\bf k}{2}\right ] \right\}f_0({\bf v}')\phi_1(\omega,{\bf k}),\ ] ] where @xmath378 is the dirac delta function .
now , the integration can be performed over @xmath643 space , obtaining the result @xmath662\phi_1(\omega,{\bf k } ) .
\end{split } \label{eq_b7}\ ] ] solving for @xmath655 in ( [ eq_b7 ] ) and inserting the result in ( [ eq_b4 ] ) , we obtain the dispersion relation @xmath663 d^3 v=0 .
\end{split}\ ] ] suitable changes of variables in the two terms now gives @xmath664 } -\frac{1}{[-\omega+{\bf k}\cdot ( { \bf u}+\frac{\hbar{\bf k}}{2 m_e } ) ] } \right]f_0({\bf u } ) d^3 u=0 , \end{split}\ ] ] which can be rewritten as @xmath665 the dispersion relation ( [ eq_b10 ] ) was also derived by bohm and pines @xcite using a series of canonical transformations of the hamiltonian of the system ; see for example equation ( 57 ) in their paper .
we now choose a coordinate system such that the @xmath299 axis is aligned with the wave vector @xmath67 .
then , ( [ eq_b10 ] ) takes the form @xmath666 we next consider a dense plasma with degenerate electrons in the zero temperature limit .
then , the background distribution function takes the simple form @xmath667 where @xmath668 is the speed of an electron on the fermi surface , and @xmath669 is the fermi energy . the integration in ( [ eq_b11 ] ) can be performed over velocity space perpendicular to @xmath670 , using cylindrical coordinate in @xmath671 and @xmath672 , obtaining the result @xmath673 where @xmath674 equation ( [ eq_b13 ] ) can be written as @xmath675 first , in the limit @xmath63 , we have from ( [ eq_b15 ] ) @xmath676 where we have assumed that @xmath65 is real and @xmath66 . expanding ( [ eq_b16 ] ) for small wavenumbers up to terms containing @xmath677
, we have @xmath678 second , for non - zero @xmath679 , we have from ( [ eq_b15 ] ) the dispersion relation @xmath62 \log\left| \frac { \frac{\omega}{k}-v_{fe}+\frac{\hbar k}{2 m_e } } { \frac{\omega}{k}+v_{fe}+\frac{\hbar k}{2 m_e } } \right| \right .
\\ \left .
+ \frac{m_e}{\hbar k v_{fe}}\left[v_{fe}^2-\left(\frac{\omega}{k}-\frac{\hbar k}{2 m_e}\right)^2\right ] \log\left| \frac { \frac{\omega}{k}-v_{fe}-\frac{\hbar k}{2 m_e } } { \frac{\omega}{k}+v_{fe}-\frac{\hbar k}{2 m_e } } \right| \right\}=0 .
\end{split } \label{eq_b18}\ ] ] expanding ( [ eq_b18 ] ) for small wavenumbers up to terms containing @xmath69 , we obtain @xmath70 where @xmath680 and @xmath72 is the bohr radius . for a typical metal such as gold , which has a free electron number density of @xmath681 , we would have @xmath682 . for the free electron density in semiconductors , which is many orders of magnitude less than in metals
, @xmath175 is much smaller and can safely be dropped compared to unity . | dense quantum plasmas are ubiquitous in planetary interiors and in compact astrophysical objects ( e.g. the interior of white dwarf stars , in magnetars etc . ) , in semiconductors and micro - mechanical systems , as well as in the next generation intense laser - solid density plasma interaction experiments and in quantum x - ray free - electron lasers .
in contrast to classical plasmas , one encounters extremely high plasma number density and low temperature in quantum plasmas .
the latter are composed of electrons , positrons and holes , which are degenerate .
positrons ( holes ) have the same ( slightly different ) mass as electrons , but opposite charge .
the degenerate charged particles ( electrons , positrons , holes ) follow the fermi - dirac statistics . in quantum plasmas , there are new forces associated with i ) quantum statistical electron and positron pressures , ii ) electron and positron tunneling through the bohm potential , and iii ) electron and positron angular momentum spin .
inclusion of these quantum forces provides possibility of very high - frequency dispersive electrostatic and electromagnetic waves ( e.g. in the hard x - ray and gamma rays regimes ) having extremely short wavelengths . in this review paper , we present theoretical backgrounds for some important nonlinear aspects of wave - wave and wave - electron interactions in dense quantum plasmas .
specifically , we shall focus on nonlinear electrostatic electron and ion plasma waves , novel aspects of 3d quantum electron fluid turbulence , as well as nonlinearly coupled intense electromagnetic waves and localized plasma wave structures .
also discussed are the phase space kinetic structures and mechanisms that can generate quasi - stationary magnetic fields in dense quantum plasmas .
the influence of the external magnetic field and the electron angular momentum spin on the electromagnetic wave dynamics is discussed .
finally , future perspectives of the nonlinear quantum plasma physics are highlighted . |
the discovery in 2012 of a new scalar particle at the lhc @xcite , whose mass has been set to @xmath13 125.09 @xmath14 0.21 ( stat . )
@xmath14 0.11 ( syst . ) gev @xcite , lays on the table the challenging issue of whether it is actually the higgs boson from the standard model of particle physics ( sm ) or there is new physics beyond the sm ( bsm ) . in this article
, we focus on one of the new physics aspects of the discovered boson the possibility of lepton flavor violating higgs decays ( lfvhd ) .
in fact , very recently , the first direct search of the particular decay @xmath15 ( from now on , we will refer to both @xmath16 and @xmath17 decays in this shortened way ) has been performed by the cms collaboration @xcite , and an upper limit on the branching ratio of br(@xmath15 ) @xmath18 at 95% c.l . has been set .
additionally , cms has also observed a slight excess with a significance of 2.4 standard deviations at @xmath13 125 gev , whose best - fit branching ratio , if interpreted as a signal , is br(@xmath15 ) @xmath19 .
the atlas collaboration has just released their results for the same @xmath15 decay @xcite as well , focusing on hadronically decaying @xmath20 leptons .
atlas has reported an upper limit of br(@xmath15 ) @xmath21 at 95% c.l . in agreement with the previous cms result .
intriguingly , a small excess appears in one of the signal regions considered , even though it is not statistically significant .
one way or another , the searches for lepton flavor violation ( lfv ) in the higgs sector have entered into the percent level .
the statistical significance is not enough to reach a strong conclusion yet , but any evidence of lfv would unquestionably mean a clear bsm signal due to the huge suppression of lfv in the sm because of the absence of flavor - changing neutral currents . in particular
, the investigation of lfvhd is at present a very active field which is being studied in different models .
lfvhd were considered for the first time in the context of the sm enlarged with three heavy majorana neutrinos in @xcite and later in the context of the type i seesaw model in @xcite , predicting tiny rates due to the strong suppression from the large heavy right - handed neutrino masses . by contrast , in the context of the inverse seesaw model ( iss ) @xcite with right - handed neutrino masses at the @xmath3 energy scale , much larger lfvhd rates , up to @xmath22 , can be obtained @xcite .
in addition , lfvhd have been also analyzed with special attention in the literature within the framework of supersymmetric ( susy ) models @xcite , finding branching ratios slightly larger than in the iss case , up to @xmath23 . here
we will study the lfvhd within the context of the susy version of the iss , which we refer to here as the susy - iss model .
in particular , we will present our estimate of the contribution to the br(@xmath17 ) from all the susy loops containing sneutrinos and sleptons which are typically different in the susy - iss with respect to other susy models , due to the important effects induced by the right - handed neutrinos and their susy partners with masses at @xmath3 .
the potential increase of the lfvhd rates due to some of the new susy loops within the susy - iss model was first pointed out and estimated in @xcite
. other important enhancement due to susy loops have also been found in @xcite for lfv lepton decay rates and other observables .
some phenomenological implications at the lhc of susy - iss scenarios with large lfvhd rates within the same context as this work have been recently studied in @xcite .
in addition to performing a complete one - loop computation of the susy loops within the susy - iss model , one of our main goals here is to analyze in detail if the enhancement due to the sneutrinos and sleptons loops can be sufficiently large as to explain the lfvhd effect seen by cms and atlas .
indeed , we will localize in this work some regions of the susy - iss parameter space where this is possible . in section [ th - framework ]
we describe the susy - iss model and introduce the parametrization we use to reproduce low - energy neutrino data . in section [ sec : analytical ] we present the analytical results of our one loop calculation while we discuss our numerical predictions in section [ results ] .
in this section , we briefly summarize the most relevant aspects for the present computation of the susy - iss model , which is a well - known extension of the mssm that can reproduce the observed neutrino masses and mixing .
the mssm superfield content is supplemented by three pairs of gauge singlet chiral superfields @xmath24 and @xmath25 with opposite lepton numbers ( @xmath26 ) .
the susy - iss model is defined by the following superpotential : @xmath27 with @xmath28 and @xmath29\,.\end{aligned}\ ] ] the generation indices have been suppressed and should be understood in a tensor notation as @xmath30 . in particular , all chiral superfields are left - handed , meaning that for @xmath31 the spin @xmath32 and spin @xmath33 components are , for example in the case of @xmath34 , @xmath35 $ ] .
@xmath36 and @xmath37 are , respectively , the down - type and up - type higgs bosons , defined as @xmath38 then the soft susy breaking lagrangian is given by @xmath39 with @xmath40 .
\nonumber\end{aligned}\ ] ] during this study we will take all soft susy breaking masses to be flavor diagonal , making sure that the only sources of flavor violation are in the neutrino yukawa coupling @xmath41 , the lepton number conserving mass term @xmath42 and the lepton number violating mass term @xmath43
. the only exception will be @xmath44 which receives rge - induced corrections coming from @xmath41 that , for phenomenological purposes , are given by @xcite @xmath45 where we take @xmath46 for the rest of this work .
after electroweak symmetry breaking , the neutrino mass matrix in the basis @xmath47 is given by @xmath48 where we have defined @xmath49 , with @xmath50 , @xmath51 and @xmath52 in order to agree with the definitions used in our previous article on lfv higgs decays @xcite . in the limit @xmath53 , it is possible to diagonalize by blocks this matrix @xcite , leading to the @xmath54 light neutrino mass matrix @xmath55 which , in turn , is diagonalized by the pmns matrix @xmath56@xcite : @xmath57 where @xmath58 is the diagonal matrix that contains the masses of the three lightest neutrinos .
low - energy neutrino data can be reproduced by using the following parametrization introduced in @xcite : @xmath59 in particular , this parametrization allows us to use the neutrino yukawa couplings @xmath41 as input parameters . as we showed in @xcite
, the following three @xmath41 textures @xmath60 where @xmath61 is a scaling factor , can lead to large @xmath62 flavor transition rates while suppressing @xmath63 and @xmath64 flavor transition rates .
we found that in the nonsupersymmetric iss model , these could lead to large branching ratios for lfv higgs decays , up to @xmath22 , while still agreeing with other experimental constraints .
as mentioned in the introduction , previous studies have demonstrated that supersymmetric contributions usually enhance the lfv rates . in particular , in the present susy - iss model ,
since we consider a seesaw scale @xmath1 not far from the electroweak scale , this low value will enhance the flavor slepton mixing due to the rge - induced radiative effects by the large neutrino yukawa couplings , and this mixing will in turn generate via the slepton loops an enhancement in the lfvhd rates . on the other hand ,
new relevant couplings appear , like @xmath65 , which for right - handed sneutrinos with @xmath66 masses may lead to new loop contributions to lfvhd that could even dominate @xcite . in light of the recent cms and atlas searches @xcite for @xmath15 ,
this calls for a new and complete evaluation of the susy contributions to this observable in the susy - iss model .
in this work , we perform a full one - loop diagrammatic computation of all relevant supersymmetric loops within the susy - iss model for @xmath67 , where @xmath68 here and from now on refers to the three neutral mssm higgs bosons , @xmath69 .
this is in contrast to the previous estimate in @xcite where an effective lagrangian description of the higgs mediated contributions to lfv processes was used , which was appropriate to capture the relevant contributions at large @xmath70 , and where the mass insertion approach was used to incorporate easily the flavor slepton mixing @xmath71 , working in the electroweak basis .
however , an expansion up to the first order in the mass insertion approximation may not be appropriate for the type of scenarios studied here , due to the large flavor - nondiagonal matrix entries considered in this work . on the other hand , we are interested also in small and moderate @xmath70 values , not just in the large @xmath70 regime , and we also wish to explore more generic soft masses for the susy particles and scan over the relevant neutrino / sneutrino parameters , mainly @xmath1 , @xmath4 and @xmath72 , not focusing only on scenarios with universal or partially universal soft parameters nor fixing the relevant parameters to one value as in @xcite . thus , our calculation is performed instead in the mass basis for all the susy particles involved in the loops , i.e , the charged sleptons , sneutrinos , charginos , and neutralinos . before moving to the calculation ,
let us introduce the relevant interaction terms from the lagrangian for the study of the lfv higgs decays .
following the notation in @xcite , these terms are given in the mass basis by @xmath73 \tilde \chi_j^- \tilde \nu_{\alpha } + h.c.\ , , \nonumber \\
\mathcal{l}_{\tilde \chi_a^0 \ell \tilde \ell_{\alpha } } & = -g\bar{\ell } \left [ b_{l \alpha a}^{(\ell ) } p_l + b_{r \alpha a}^{(\ell ) } p_r \right ] \tilde \chi_a^0 \tilde \ell_{\alpha } + h.c.\ , , \nonumber \\ \mathcal{l}_{h_x \tilde s_{\alpha } \tilde s_{\beta}}&= -\imath h_x \left [ g_{h_x\tilde \nu_{\alpha } \tilde \nu_{\beta } } \tilde \nu_{\alpha}^ * \tilde \nu_{\beta } + g_{h_x\tilde \ell_{\alpha } \tilde \ell_{\beta } } \tilde \ell_{\alpha}^ * \tilde \ell_{\beta } \right]\,,\nonumber\\ \mathcal{l}_{h_x \tilde \chi_i^- \tilde \chi_j^-}&= - g h_x\bar{\tilde{\chi}}_i^- \left [ w_{lij}^{(x)}p_l+ w_{rij}^{(x)}p_r \right ] \tilde{\chi}_j^-\,,\nonumber\\ \mathcal{l}_{h_x \tilde \chi_a^0 \tilde \chi_b^0}&= - \frac{g}{2 } h_x\bar{\tilde { \chi}}_a^0 \left [ d_{lab}^{(x)}p_l+ d_{rab}^{(x)}p_r \right ] \tilde { \chi}_b^0\ , , \nonumber \\ \mathcal{l}_{h_x\ell\ell } & = -g h_x\bar{\ell } \left [ s_{l,\ell}^{(x ) } p_l + s_{r,\ell}^{(x ) } p_r \right ] \ell \,,\end{aligned}\ ] ] where the coupling factors have been expressed in terms of the susy - iss model parameters and are collected in appendix [ app_couplings ] .
we take into account the full set of 1-loop susy diagrams shown in figure [ susy - diag ] .
it is interesting to notice that , since we work in the mass basis , the set of diagrams contributing to lfv higgs decays ( four diagrams with charginos and sneutrinos in the loops , and four more with neutralinos and charged sleptons ) is the same as in the susy type i seesaw model which was considered in @xcite .
we keep their definition of the form factors @xmath74 where @xmath75 is the decay amplitude for @xmath76 with again @xmath69 and @xmath77 is the ingoing higgs boson momentum .
the contributions of the susy diagrams are summed in @xmath78 and @xmath79 according to @xmath80 their analytic expressions are taken from @xcite and reproduced in appendix [ app_ff ] for completeness , including the proper modifications to adapt them to the susy - iss model .
we have checked analytically the cancellation of divergences appearing in the loop contributions in both form factors @xmath78 and @xmath79 in eq.(16),giving , as expected , a finite result without the need of the renormalization procedure .
notice that this cancellation is not trivial and is , therefore , a good test of our results of the form factors in appendix [ app_ff ] .
the parametrization of the lfvhd widths in terms of form factors remains unchanged and is given by @xmath81 where @xmath82 is the @xmath83 coupling constant , @xmath84 is the mass of the higgs boson while @xmath85 and @xmath86 are the masses of final state leptons .
in this section we show the numerical results of the lfv decay rates of the lightest neutral higgs boson , br(@xmath87 ) , as a function of the most relevant parameters of the susy - iss model for the full susy contribution to lfvhd , namely , @xmath1 , @xmath4 and @xmath72 .
it should be noted that , in the absence of cp violation , as in our case , br(@xmath88)=br(@xmath89 ) and , therefore , in comparing with data the two rates should be added .
we have imposed various experimental constraints , choosing as example two benchmark points leading to a higgs boson mass within 1@xmath90 of the central value of the latest cms and atlas combination , and with supersymmetric spectrum allowed by atlas and cms searches .
indeed , we concentrate for this work on the slepton sector , since the squark sector is irrelevant for the lfvhd .
the squark parameters are only relevant for the higgs mass prediction , thus one can always adjust the squark masses and the trilinear couplings @xmath91 and @xmath92 in order to ensure a correct higgs boson mass .
we also restrict ourselves to the case of @xmath93 , avoiding constraints from the invisible higgs decay widths , and consider only real @xmath94 and mass matrices , making constraints from lepton electric dipole moments irrelevant .
finally , we also take into account the lfv radiative decays whose current upper limits at the @xmath95 c.l .
are @xmath96 points excluded by lfv radiative decays will be denoted by a cross , while a triangle will represent the ones allowed .
we present here the predictions of br(@xmath17 ) for the three neutrino yukawa textures exposed in the section [ th - framework ] , ensuring the practically vanishing lfv in the @xmath63 sector , i.e. , leading to br@xmath97 and br@xmath98 .
it should be noticed that these textures also suppress substantially the lfv in the @xmath99 sector .
therefore , the most stringent constraint , making use of these textures , is that of the related lfv radiative decay @xmath100 . in figure
[ lfvhd - mr ] , [ cols="^,^ " , ] where these parameters are varied independently from the susy scale .
as when varying @xmath1 , increasing @xmath2 and @xmath101 makes the singlet sneutrinos heavier and decreases the size of the chargino contribution . for @xmath102 and @xmath103 which are dominated by this contribution , the br(@xmath17 )
exhibits a strong decrease between 200 gev and 14 tev , by more than five orders of magnitude in the case of @xmath103 . for @xmath104
a dip can be observed , due again to a cancellation between the chargino and neutralino contributions , with the latter dominating at large @xmath2 .
for the first benchmark point , the largest @xmath17 rates allowed by the @xmath100 upper limit are obtained for @xmath103 with @xmath105 200 gev , with a maximum value of @xmath106 , just one order of magnitude below the present lhc sensitivity .
if we move our attention to the vicinity of the region of low values of @xmath2 for the second benchmark point , we found large lfvhd rates , as displayed on the right panel of figure [ lfvhd - msoft ] , with @xmath107 200 gev and @xmath108 2.5 tev .
we observe a huge increase in br(@xmath17 ) for the three yukawa textures @xmath102 , @xmath103 , and @xmath104 , with maximum values of @xmath109 , @xmath110 , and @xmath111 , respectively , due mainly to the low values of @xmath2 and @xmath1 .
unfortunately , all the parameter space for @xmath102 and @xmath103 cases is excluded by the @xmath100 upper limit . by contrast , most of the points for the @xmath104 texture are in agreement with this upper bound , because they are located in a region where the @xmath100 rates suffer a strong suppression as a consequence of the value set for @xmath4 in this case , @xmath108 2.5 tev .
this fact allows us to obtain a maximum value of br(@xmath17 ) @xmath112 , completely within the reach of the current lhc experiments and large enough to explain the cms and atlas excesses if confirmed by other experiments and/or future data .
in this article , we have presented the results of an updated and full one - loop calculation of the susy contributions to lepton flavor violating higgs decays in the susy - iss model .
we found much larger contributions than in the type i seesaw due to the lower values of @xmath113 tev ) , an increased rge - induced slepton mixing , and the presence of right - handed sneutrinos at the tev scale , where both sleptons and sneutrinos large couplings transmit sizable lfv due to the large @xmath114 considered here .
we showed that the branching ratio of @xmath87 exhibits different behaviors as a function of the seesaw and susy scale if it is dominated by chargino or neutralino loops .
moreover , a nonzero trilinear coupling @xmath4 leads to increased lfvhd rates . choosing different benchmark points
, we found that br(@xmath115 ) of the order of @xmath116 can be reached while agreeing with the experimental limits on radiative decays , providing a possible explanation of the cms and atlas excesses . while out of the scope of this work , a complete study including nonsupersymmetric contributions in the susy - iss model and a detailed analysis of experimental constraints beyond radiative lfv decays will be presented in a future article .
this work is supported by the european union grant no .
fp7 itn invisibles ( marie curie actions , grant no .
pitn - ga-2011 - 289442 ) , by the cicyt through grant no .
fpa2012 - 31880 , by the spanish consolider - ingenio 2010 programme cpan ( grant no . csd2007 - 00042 ) , and by the spanish mineco s `` centro de excelencia severo ochoa '' programme under grant no .
sev-2012 - 0249 .
e. a. is financially supported by the spanish dgiid - dga grant no .
2013-e24/2 and the spanish micinn grants no . fpa2012 - 35453 and no .
cpan - csd2007 - 00042 .
x. m. is supported through the fpu grant no .
ap-2012 - 6708 .
c. w. received financial support as an international research fellow of the japan society for the promotion of science and from the european research council under the european union s seventh framework programme ( fp/2007 - 2013 ) / erc grant numass agreement no .
[ 617143 ] during different stages of this work .
we present in this appendix the mass matrices and coupling factors that are relevant to our calculation of the lfv higgs decays . the sneutrino mass matrix @xmath117 is defined by @xmath118 where @xmath119 , @xmath120 and @xmath121 are vectors made of weak eigenstates and defined in a similar fashion , e.g. @xmath122 .
the @xmath123 sneutrino mass matrix is expressed in terms of @xmath124 submatrices , giving @xmath125 with @xmath126 where we have used the fact that @xmath127 is a symmetric matrix .
then , the sneutrino mass matrix is diagonalized using @xmath128 which corresponds to @xmath129 the basis in eq .
( [ msnu ] ) uses the sneutrino electroweak eigenstates and their complex conjugate states , and they fulfill @xmath130 and @xmath131 since the physical sneutrinos are real scalar fields . while both eqs .
( [ thisone ] ) and ( [ thistwo ] ) are equally valid , we choose eq .
( [ thisone ] ) .
the mass matrices of the other susy particles , namely the charginos , neutralinos , and charged sleptons , are the same as in the susy type i seesaw studied in @xcite and we will use their definitions of the corresponding rotation matrices , which in turns were based on the conventions of @xcite for the charginos and neutralinos . concretely , @xmath132 and @xmath133 will be the matrices that rotate the chargino states and @xmath134 the one that rotates the neutralino states .
in addition , combinations of rotation matrices for the neutralinos are defined as @xmath135 as for the charged sleptons , they are diagonalized by @xmath136 where @xmath137 are the weak eigenstates and @xmath138 are the mass eigenstates . when compared with the susy type i seesaw , only the coupling factors @xmath139 and @xmath140 are modified . in the susy inverse seesaw , they are defined in the mass basis with diagonal charged leptons by @xmath141\ , , \nonumber\\ ( g_{ll , \nu}^{(x)})_{ik } = & - \frac{m_z}{2\cos{\theta_w } } \sigma_3^{(x ) } \delta_{ik } + \frac{(m_d m_d^\dagger)_{ik}}{m_w \sin \beta } \sigma_6^{(x)}\ , , \nonumber \\
( g_{rr , \nu}^{(x)})_{ik } = & \frac{(m_d^\dagger m_d)_{ik}}{m_w \sin\beta } \sigma_6^{(x)}\ , , \nonumber \\ ( g_{lr , \nu}^{(x)})_{ik } = & \frac{(m_d a_\nu^\dagger)_{ik}}{2m_w\sin\beta } \sigma_2^{(x ) } + \frac{\mu } { 2m_w\sin \beta } ( m_d)_{ik } \sigma_7^{(x ) } \ , , \nonumber \\ ( g_{lx , \nu}^{(x)})_{ik } = & \frac{(m_d m_r^*)_{ik}}{2 m_w \sin\beta } \sigma_2^{(x)}\,,\end{aligned}\ ] ] which are summed over the internal indices , with @xmath142 .
we reproduced below the unmodified coupling factors from @xcite ( correcting a typo in @xmath143 ) for completeness in the mass basis with diagonal charged leptons @xmath144r_{(2,4,6)\alpha}^{(\ell)}\right]\ , , \nonumber \\
b_{r\alpha a}^{(e,\mu,\tau)}= & \sqrt{2}\left[\left(-\sin\theta_wn_{a1}^{'}-\frac{1}{\cos\theta_w}(\frac{1}{2}-\sin^2\theta_w)n_{a2}^{'}\right ) r_{(1,3,5)\alpha}^{(\ell)}+ \frac{m_{e,\mu,\tau}}{2m_w \cos\beta}n_{a3}r_{(2,4,6)\alpha}^{(\ell)}\right]\ , , \nonumber \\
w_{lij}^{(x)}=&\frac{1}{\sqrt{2}}\left(-\sigma_1^{(x)}u_{j2}^*v_{i1}^*+ \sigma_2^{(x)}u_{j1}^*v_{i2}^*\right)\ , , \nonumber \\
w_{rij}^{(x)}=&\frac{1}{\sqrt{2}}\left(-\sigma_1^{(x)*}u_{i2}v_{j1}+\sigma_2^{(x)*}u_{i1}v_{j2}\right)\ , , \nonumber \\
d_{lab}^{(x)}=&\frac{1}{2\cos\theta_w}\left[(\sin\theta_w n_{b1}^*-\cos\theta_w n_{b2}^*)(\sigma_1^{(x)}n_{a3}^*+\sigma_2^{(x)}n_{a4}^ * ) \right .
\nonumber \\ & + ( \sin\theta_w n_{a1}^*-\cos\theta_w n_{a2}^*)(\sigma_1^{(x)}n_{b3}^*+\sigma_2^{(x)}n_{b4}^*)\left.\right ] \ , , \nonumber \\ d_{rab}^{(x)}=&d_{lab}^{(x)*}\ , , \nonumber \\
s_{l,\ell}^{(x ) } = & - \frac{m_{\ell}}{2 m_w \cos\beta}{\sigma_1^{(x)*}}\ , , \nonumber \\
s_{r,\ell}^{(x ) } = & s_{l , \ell}^{(x ) * } \ , , \nonumber
\\ g_{h_x \tilde{\ell}_{\alpha } \tilde{\ell}_{\beta } } = & -\imath g \left [ g_{ll , e}^{(x ) } r_{1\alpha}^{*(\ell ) } r_{1\beta}^{(\ell ) } + g_{rr , e}^{(x ) } r_{2\alpha}^{*(\ell ) } r_{2\beta}^{(\ell ) } + g_{lr , e}^{(x ) } r_{1\alpha}^{*(\ell ) } r_{2\beta}^{(\ell ) } + g_{rl , e}^{(x ) } r_{2\alpha}^{*(\ell ) } r_{1\beta}^{(\ell)}\right .
\nonumber \\ & + g_{ll , \mu}^{(x ) } r_{3\alpha}^{*(\ell ) } r_{3\beta}^{(\ell ) } + g_{rr , \mu}^{(x ) } r_{4\alpha}^{*(\ell ) } r_{4\beta}^{(\ell ) } + g_{lr , \mu}^{(x ) } r_{3\alpha}^{*(\ell ) } r_{4\beta}^{(\ell ) } + g_{rl , \mu}^{(x ) } r_{4\alpha}^{*(\ell ) } r_{3\beta}^{(\ell ) } \nonumber \\ & + \left .
g_{ll , \tau}^{(x ) } r_{5\alpha}^{*(\ell ) } r_{5\beta}^{(\ell ) } + g_{rr , \tau}^{(x ) } r_{6\alpha}^{*(\ell ) } r_{6\beta}^{(\ell ) } + g_{lr , \tau}^{(x ) } r_{5\alpha}^{*(\ell ) } r_{6\beta}^{(\ell ) } + g_{rl , \tau}^{(x ) } r_{6\alpha}^{*(\ell ) } r_{5\beta}^{(\ell ) } \right ] \ , , \nonumber \\
g_{ll , \ell}^{(x ) } = & \frac{m_z}{\cos{\theta_w } } \sigma_3^{(x ) } \left ( \frac{1}{2}- \sin^2{\theta_w } \right ) + \frac{m_{\ell}^2}{m_w \cos{\beta } } \sigma_4^{(x)}\ , , \nonumber\\ g_{rr , \ell}^{(x ) } = & \frac{m_z}{\cos{\theta_w } } \sigma_3^{(x ) } \left ( \sin^2{\theta_w } \right ) + \frac{m_{\ell}^2}{m_w \cos{\beta } } \sigma_4^{(x)}\,,\nonumber \\ g_{lr , \ell}^{(x ) } = & \left(-\sigma_1^{(x)}a_\ell-\sigma_5^{(x)}\mu\right ) \frac{m_{\ell}}{2 m_w \cos{\beta}}\ , , \nonumber \\
g_{rl , \ell}^{(x ) } = & g_{lr , \ell}^{(x)*}\,,\end{aligned}\ ] ] with @xmath145
we present here the form factors that correspond to the diagrams of figure [ susy - diag ] .
the original calculation in the susy type i seesaw was carried by some of the authors in the mass basis and in the feynmant hooft gauge @xcite .
the only changes required to adapt the original form factors to the susy - iss model are the sum over sneutrinos that has to be extended to the 18 mass eigenstates and the new couplings defined in appendix [ app_couplings ] . in the following formulas , summation over all indices corresponding to internal propagators is understood .
these would be @xmath146 for the sneutrinos , @xmath147 for the charginos , @xmath148 for the charged sleptons and @xmath149 for the neutralinos .
@xmath150\ , , \nonumber \\
f_{l , x}^{(2 ) } & = & - \frac{igg_{h_x\tilde { \nu}_\alpha \tilde { \nu}_\beta}}{16\pi^2 } \left[-m_{\ell_k}(c_{11}-c_{12})\,\iota_{l 1}^{x,\tilde \chi^-}- m_{\ell_m } c_{12}\ , \iota_{l 2}^{x,\tilde \chi^-}+ m_{\tilde \chi^-_i}c_0\,\iota_{l 3}^{x,\tilde \chi^-}\right]\,,\nonumber \\ f_{l , x}^{(3 ) } & = & - \frac{s_{l,\ell_{m}}^{(x)}}{m_{\ell_k}^2-m_{\ell_m}^2}\left[m_{\ell_k}^2 \sigma_r^{\tilde \chi^- } ( m_{\ell_k}^2)+m_{\ell_k}^2 \sigma_{rs}^{\tilde \chi^-}(m_{\ell_k}^2 ) + m_{\ell_m}\left(m_{\ell_k } \sigma_l^{\tilde \chi^- } ( m_{\ell_k}^2)+ m_{\ell_k}\sigma_{ls}^{\tilde \chi^- } ( m_{\ell_k}^2)\right)\right]\,,\nonumber \\ f_{l , x}^{(4 ) } & = & - \frac{s_{l,\ell_{k}}^{(x)}}{m_{\ell_m}^2-m_{\ell_k}^2 } \left[m_{\ell_m}^2 \sigma_l^{\tilde \chi^- } ( m_{\ell_m}^2)+ m_{\ell_m } m_{\ell_k } \sigma_{rs}^{\tilde \chi^-}(m_{\ell_m}^2)\right
. + \left .
m_{\ell_k}\left(m_{\ell_m } \sigma_r^{\tilde \chi^- } ( m_{\ell_m}^2 ) + m_{\ell_k } \sigma_{ls}^{\tilde \chi^- } ( m_{\ell_m}^2)\right)\right]\,,\nonumber \\ f_{l , x}^{(5 ) } & = & - \frac{g^2}{16 \pi^2}\left[\left(b_0 + m_{\tilde \ell_{\alpha}}^2 c_0+m_{\ell_m}^2 c_{12}+m_{\ell_k}^2 ( c_{11 } - c_{12})\right)\ , \kappa_{l 1}^{x,\,\tilde \chi^0 } \right.\nonumber \\ & & + m_{\ell_k } m_{\ell_m } \left(c_{11}+c_0\right)\kappa_{l 2}^{x , \tilde \chi^0}+ m_{\ell_k } m_{\tilde \chi _ b^0 } \left(c_{11}-c_{12}+c_0\right)\kappa_{l 3}^{x , \tilde \chi^0}\,+ m_{\ell_m } m_{\tilde \chi _ b^0 } c_{12}\,\kappa_{l 4}^{x,\tilde \chi^0}\nonumber \\ & & + m_{\ell_k } m_{\tilde \chi _ a^0 } \left(c_{11}-c_{12}\right)\kappa_{l 5}^{x,\tilde \chi^0 } + m_{\ell_m } m_{\tilde \chi _ a^0 } \left(c_{12}+c_0\right)\kappa_{l 6}^{x,\tilde \chi^0}+ m_{\tilde \chi _ a^0 } m_{\tilde \chi _ b^0}c_0\,\kappa_{l 7}^{x,\tilde \chi^0 } \big]\ , , \nonumber \\
f_{l , x}^{(6 ) } & = & - \frac{igg_{h_x\tilde \ell_\alpha \tilde \ell_\beta}}{16\pi^2 } \left[-m_{\ell_k}(c_{11}-c_{12})\,\iota_{l 1}^{x,\tilde \chi^0}- m_{\ell_m } c_{12}\ , \iota_{l 2}^{x,\tilde \chi^0}+ m_{\tilde \chi_a^0}c_0\,\iota_{l 3}^{x,\tilde \chi^0}\right]\,,\nonumber \\ f_{l , x}^{(7 ) } & = & - \frac{s_{l,\ell_m}^{(x)}}{m_{\ell_k}^2-m_{\ell_m}^2}\left[m_{\ell_k}^2 \sigma_r^{\tilde \chi^0 } ( m_{\ell_k}^2)+m_{\ell_k}^2 \sigma_{rs}^{\tilde \chi^0}(m_{\ell_k}^2 ) + m_{\ell_m}\left(m_{\ell_k } \sigma_l^{\tilde \chi^0 } ( m_{\ell_k}^2)+ m_{\ell_k}\sigma_{ls}^{\tilde \chi^0 } ( m_{\ell_k}^2)\right)\right]\,,\nonumber \\ f_{l , x}^{(8 ) } & = & - \frac{s_{l,\ell_k}^{(x)}}{m_{\ell_m}^2-m_{\ell_k}^2 } \left[m_{\ell_m}^2 \sigma_l^{\tilde \chi^0 } ( m_{\ell_m}^2)+ m_{\ell_m } m_{\ell_k } \sigma_{rs}^{\tilde \chi^0}(m_{\ell_m}^2)\right
. + \left .
m_{\ell_k}\left(m_{\ell_m } \sigma_r^{\tilde \chi^0 } ( m_{\ell_m}^2 ) + m_{\ell_k } \sigma_{ls}^{\tilde \chi^0 } ( m_{\ell_m}^2)\right)\right]\,,\nonumber \label{formfactorlbs}\end{aligned}\ ] ] where , @xmath151 and @xmath152 the couplings and self - energies from the neutralino contributions to the form factors were defined as @xmath153 @xmath154 the couplings and self - energies from the chargino contributions to the form factors , @xmath155 , @xmath156 , and @xmath157 can be obtained from the previous expressions @xmath158 , @xmath159 and @xmath160 by using the following replacement rules @xmath161 , @xmath162 , @xmath163 , @xmath164 , @xmath165 , and @xmath166 .
g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * ( 2012 ) 1 [ arxiv:1207.7214 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * ( 2012 ) 30 [ arxiv:1207.7235 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas and cms collaborations ] , phys . rev .
* 114 * ( 2015 ) 191803 [ arxiv:1503.07589 [ hep - ex ] ] .
v. khachatryan _ et al . _
[ cms collaboration ] , phys .
b * 749 * ( 2015 ) 337 [ arxiv:1502.07400 [ hep - ex ] ] .
g. aad _ et al .
_ [ atlas collaboration ] , jhep * 1511 * ( 2015 ) 211 doi:10.1007/jhep11(2015)211 [ arxiv:1508.03372 [ hep - ex ] ] .
a. pilaftsis , phys .
b * 285 * ( 1992 ) 68 .
e. arganda , a. m. curiel , m. j. herrero and d. temes , phys .
d * 71 * ( 2005 ) 035011 [ hep - ph/0407302 ] .
r. n. mohapatra , phys .
* 56 * ( 1986 ) 561 ; r. n. mohapatra and j. w. f. valle , phys .
d * 34 * ( 1986 ) 1642 ; j. bernabeu , a. santamaria , j. vidal , a. mendez and j. w. f. valle , phys .
b * 187 * ( 1987 ) 303 .
e. arganda , m. j. herrero , x. marcano and c. weiland , phys .
d * 91 * ( 2015 ) 015001 [ arxiv:1405.4300 [ hep - ph ] ] .
see , for example , j. l. diaz - cruz , jhep * 0305 * ( 2003 ) 036 [ hep - ph/0207030 ] ; a. brignole and a. rossi , phys .
b * 566 * ( 2003 ) 217 [ hep - ph/0304081 ] ; a. brignole and a. rossi , nucl .
b * 701 * ( 2004 ) 3 [ hep - ph/0404211 ] ; a. vicente , adv . high energy phys .
* 2015 * ( 2015 ) 686572 [ arxiv:1503.08622 [ hep - ph ] ] .
m. arana - catania , e. arganda and m. j. herrero , jhep * 1309 * ( 2013 ) 160 [ jhep * 1510 * ( 2015 ) 192 ] [ arxiv:1304.3371 [ hep - ph ] ] .
a. abada , d. das and c. weiland , jhep * 1203 * ( 2012 ) 100 [ arxiv:1111.5836 [ hep - ph ] ] .
a. abada , m. e. krauss , w. porod , f. staub , a. vicente and c. weiland , jhep * 1411 * ( 2014 ) 048 [ arxiv:1408.0138 [ hep - ph ] ] .
e. arganda , m. j. herrero , x. marcano and c. weiland , phys .
b * 752 * ( 2016 ) 46 doi:10.1016/j.physletb.2015.11.013 [ arxiv:1508.05074 [ hep - ph ] ] .
j. hisano , t. moroi , k. tobe and m. yamaguchi , phys .
d * 53 * ( 1996 ) 2442 [ hep - ph/9510309 ] .
m. c. gonzalez - garcia and j. w. f. valle , phys .
b * 216 * ( 1989 ) 360 .
b. pontecorvo , sov .
jetp * 6 * ( 1957 ) 429 [ zh . eksp .
fiz . * 33 * ( 1957 ) 549 ] .
z. maki , m. nakagawa and s. sakata , prog .
* 28 * ( 1962 ) 870 .
j. adam _ et al .
_ [ meg collaboration ] , phys . rev .
* 110 * ( 2013 ) 201801
[ arxiv:1303.0754 [ hep - ex ] ] . b. aubert _ et al . _
[ babar collaboration ] , phys .
* 104 * ( 2010 ) 021802 [ arxiv:0908.2381 [ hep - ex ] ] . j. f. gunion and h. e. haber , nucl .
b * 272 * ( 1986 ) 1 [ nucl .
b * 402 * ( 1993 ) 567 ] . | in this article we study the full one - loop susy contributions to the lepton flavor violating higgs decay @xmath0 , within the context of the supersymmetric inverse sesaw model .
we assume that both the right - handed neutrino masses , @xmath1 , and their supersymmetric partner masses , @xmath2 , are not far from the interesting @xmath3 energy scale , and we work with scenarios with large neutrino yukawa couplings that transmit large lepton flavor violating effects . by exploring the behavior with the most relevant parameters , mainly @xmath1 , @xmath2 and the trilinear sneutrino coupling @xmath4
, we will look for regions of the parameter space where the enhancement of @xmath5 is large enough to reach values at the percent level , which could explain the excess recently reported by cms and atlas at the cern large hadron collider .
ift - uam / csic-15 - 085 + ftuam-15 - 25 + su - het-07 - 2015 + ippp/15/67 + dcpt/15/134 * enhancement of the lepton flavor violating higgs boson decay rates from susy loops in the inverse seesaw model *
e. arganda@xmath6 , m.j .
herrero@xmath7 , x. marcano@xmath7 , c. weiland@xmath8 _ @xmath9departamento de fsica terica , facultad de ciencias , + universidad de zaragoza , e-50009 zaragoza , spain _
@xmath10departamento de fsica terica and instituto de fsica terica , ift - uam / csic , + universidad autnoma de madrid , cantoblanco , 28049 madrid , spain @xmath11graduate school of science and engineering , + shimane university , matsue , 690 - 8504 japan @xmath12institute for particle physics phenomenology , + department of physics , durham university , durham dh1 3le , united kingdom 2 |
to confirm that the band gaps induced by os and cu - os stem from spin - orbit coupling , we performed dft calculations for these adatoms at 6.25% coverage in the non - relativistic limit . for os / graphene , we find that in the absence of spin - orbit coupling the os spins polarize much more strongly compared to the relativistic case yielding a magnetic moment @xmath71 .
the spin splitting is clearly visible in the partial density of states for the os @xmath2 orbitals shown in fig .
[ figs - band - nosoc](a ) , where positive and negative values on the horizontal axis correspond respectively to the majority and minority spin channels .
the band structure for the majority spins ( solid curves ) and minority spins ( dashed curves ) appears in fig .
[ figs - band - nosoc](b ) .
notice that in sharp contrast to the spin - orbit - coupled case , both spin channels are always gapless for any value of the chemical potential ( at least over the energy range shown ) .
interestingly , ( cu - os)/graphene remains non - magnetic even in the absence of spin - orbit coupling . from the band structure in fig .
[ figs - band - nosoc](c ) one clearly sees that here , too , the system remains gapless in the non - relativistic limit .
thus for both os / graphene and ( cu - os)/graphene the gaps indeed originate solely from spin - orbit interactions . in the main text we remarked that cu and os co - adsorbates energetically prefer to form vertical dimers over the h - site in graphene . here
we summarize the evidence supporting this conclusion .
we used dft , without spin - orbit coupling , to explore the energetics of the various configurations displayed in fig .
[ figs - structure ] ( os , cu , and c atoms are respectively represented by dark cyan , coral , and gray spheres ) .
note that in all cases we fix the os to the h - site since the binding energy for that position greatly exceeds that for the top or bridge sites .
( additionally , cu - os dimers bind much more weakly to the top and bridge positions . )
the vertical cu - os dimer in configuration i exhibits a very large binding energy @xmath72 ev .
in contrast the binding energy for configurations iii and iv are substantially lower by 2.50 and 2.09 ev , respectively .
configuration ii is unstable and transforms to configuration i after relaxation , indicating a strong tendency toward vertical dimer formation . as we saw earlier
the vertical dimer in ( a ) is nonmagnetic ( with or without spin - orbit coupling ) , whereas the other configurations are magnetic .
finally , we briefly highlight our results for ir adatoms on graphene .
we found that ir slightly prefers the h site over the bridge site , the binding energy being 2.17 ev in the former configuration and 2.12 ev in the latter .
the binding energy for ir at the top site is weakest at 1.94 ev .
all results discussed henceforth thus correspond to h - site ir adatoms . without spin - orbit coupling dft
predicts that ir / graphene is non - magnetic .
figures [ figs - ir - band](a ) and ( b ) illustrate the band structure and partial density of states for the ir @xmath2 levels , calculated at 6.25% coverage . just as for os and cu - os dimers , here too the system remains gapless in the non - relativistic limit .
figure [ figs - ir - band](d ) shows that restoring spin - orbit coupling introduces a huge band gap @xmath69 ev below the fermi level ( green dashed line ) .
moreover , in fig .
[ figs - ir - band](c ) we see that this gap results from hybridization between the ir @xmath6 orbitals and graphene s @xmath48 bands , similar to os / graphene and the tight - binding model described in the main text .
spin - orbit interactions also , however , produce small spin and orbital moments for ir given respectively by 0.30 @xmath56 and 0.13 @xmath56 ; these are responsible for the weak band splittings in fig .
[ figs - ir - band](d ) .
coverage dependence of the gap is illustrated in fig .
[ figs - ir - coverage ] . as in all other cases studied here ,
the gap remains quite large down to very dilute ir concentrations . even at 1% coverage where @xmath55 reduces to 0.08 ev the gap exceeds that induced by in or tl adatoms @xcite many times over
observing a true topological insulator phase in ir / graphene is complicated by the magnetic moments predicted by dft and the fact that the fermi level resides in the conduction band .
( one should again keep in mind , however , that dft may overestimate the tendency for moment formation here .
furthermore , it may be feasible to gate the system back to the insulating regime by conventional means at low coverages . ) in the main text we described how co - doping with cu eliminates both potential challenges . here
we simply wish to note that cu - ir dimers carry yet another advantage
they strongly enhance binding to the h - site compared to isolated ir adatoms , similar to the os case .
indeed , the binding energy for cu - ir dimers over the h - site is larger by 0.62 ev and 0.79 ev compared to the bridge and top sites , respectively . by contrast the differences for pure ir are only 0.05 ev and 0.23 ev as we saw above . | two - dimensional topological insulators ( 2d tis ) have been proposed as platforms for many intriguing applications , ranging from spintronics to topological quantum information processing . realizing this potential will likely be facilitated by the discovery of new , easily manufactured materials in this class . with this goal in mind
we introduce a new framework for engineering a 2d ti by hybridizing graphene with impurity bands arising from heavy adatoms possessing partially filled @xmath0-shells , in particular osmium and iridium .
first principles calculations predict that the gaps generated by this means exceed 0.2 ev over a broad range of adatom coverage ; moreover , tuning of the fermi level is not required to enter the ti state .
the mechanism at work is expected to be rather general and may open the door to designing new ti phases in many materials .
topological insulators comprise a class of strongly spin - orbit - coupled , non - magnetic materials that are electrically inert in the bulk yet possess protected metallic states at their boundary @xcite .
these systems are promising sources for a host of exotic phenomena including majorana fermions @xcite , charge fractionalization @xcite , and novel magneto - electric effects @xcite and may also find use for quantum computing @xcite and spintronics devices @xcite . in some respects two - dimensional ( 2d ) topological insulators are ideally suited for such applications ; for example , bulk carriers that often plague their three - dimensional counterparts can be vacated simply by gating .
experimental progress on 2d topological insulators has steadily advanced recently due largely to pioneering work on hgte @xcite ( see also ref .
@xcite ) . nevertheless , to realize their full potential systems more amenable to experimental investigations are highly desirable . in this
regard the ability to design new 2d topological insulators from conventional , widely available materials would constitute a major step forward , and many proposals of this spirit now exist @xcite . following this strategy , here we introduce a new mechanism for engineering a topological insulator state in graphene arguably now the most broadly accessible 2d electron system . historically , graphene was the first material predicted to realize a topological insulator in seminal work by kane and mele @xcite , though unfortunately the gap is unobservably small due to carbon s exceedingly weak spin - orbit coupling @xcite .
reference @xcite revived graphene as a viable topological insulator candidate by predicting that dilute concentrations of heavy in or tl adatoms dramatically enhance the gap to detectable values of order 0.01 ev . essentially , these adatoms mediate enhanced spin - orbit interactions of the type present in the kane - mele model @xcite for pure graphene .
our approach here also relies on hybridizing graphene with dilute heavy adatoms , though the underlying physics is entirely different and can not be understood in terms of an effective graphene - only model .
rather , we will show using density functional theory that certain adatoms specifically os , ir , cu - os dimers , and cu - ir dimers form spin - orbit - split impurity bands that hybridize with graphene s dirac states in such a way that a highly robust topological insulator regime emerges .
in fact here it is more appropriate to view the adatoms as the dominant low - energy degrees of freedom , with _ their _ coupling effectively mediated by graphene ; from this perspective this mechanism represents the inverse of that invoked in ref .
@xcite .
numerous practical advantages arise in this scheme .
the topological insulator gaps are extremely large typically exceeding 0.2 ev and take on nearly the full _ atomic _ spin - orbit splitting for the adatoms .
such values reflect more than an order - of - magnitude enhancement compared to the gaps induced by in or tl , and are competitive with the largest gap predicted for any topological insulator .
somewhat counterintuitively , these gaps are remarkably insensitive to the adatom concentration , taking on comparable values at least over coverages ranging from @xmath1% . in the case of os adatoms and cu - ir dimers
, the fermi level also naturally resides within the topological insulator band gap .
this eliminates perhaps the most serious challenge with in and tl , both of which substantially electron - dope graphene even at quite low coverages .
these features suggest that the observation of a topological insulator state in graphene may be within reach .
we first elucidate the mechanism uncovered here using a tight - binding model that exposes the physics in a very transparent manner . consider @xmath2 adatoms residing at positions @xmath3 located at ` hollow ' ( h ) sites in graphene as in fig .
[ bandstructurefig](a ) .
for simplicity we retain only the @xmath4 and @xmath5 adatom states since these comprise the most important orbitals in our first - principles calculations .
( recall that @xmath6 orbitals arise from @xmath7 orbital angular momentum @xmath8 states . )
we then model the composite system by a hamiltonian @xmath9 @xcite .
the first term allows nearest - neighbor hopping for graphene : @xmath10 where @xmath11 adds an electron with spin @xmath12 to honeycomb site @xmath13 .
the second encodes couplings for the adatoms , @xmath14}. \label{ha}\end{aligned}\ ] ] here @xmath15 fills the adatom @xmath0-orbital at position @xmath3 with magnetic quantum number @xmath8 and spin @xmath12 , @xmath16 sets the orbital energies relative to graphene s dirac points , @xmath17 represents spin - orbit coupling , and @xmath18 is a pauli matrix .
finally , @xmath19 hybridizes the adatoms with graphene . to express this term
it is convenient to define vectors @xmath20 that point from an adatom to the six surrounding carbon sites [ see fig . [ bandstructurefig](a ) ] .
one can then construct linear combinations @xmath21 that carry angular momentum @xmath22 and write @xcite @xmath23 let us now specialize to a periodic adatom arrangement characterized by the @xmath24 supercell shown in fig .
[ bandstructurefig](a ) , with one adatom per cell ( this corresponds to @xmath25 coverage ) .
figure [ bandstructurefig](b ) illustrates the band structure with @xmath26 ev , @xmath27 ev , and @xmath28 . in this limit
the adatoms produce a four - fold degenerate flat band , reflecting spin and orbital degeneracy .
for the following discussion the precise location of these adatom states is unimportant , provided they intersect the carbon bands within @xmath29 ev of the dirac points . incorporating tunneling between the adatoms and graphene causes the flat bands to disperse , as shown in fig .
[ bandstructurefig](c ) for @xmath30 ev .
suppose now that the fermi level resides at the dashed line in fig .
[ bandstructurefig](c ) .
although the spectrum here exhibits a sizable energy gap near the @xmath31 point , the system remains metallic due to band touchings at the zone center .
the gapless excitations at zero momentum exhibit the following two crucial properties : @xmath32 they arise from _ weakly perturbed adatom orbitals _ since at the zone center the nearest carbon bands reside well over 1 ev away in energy , and @xmath33 they are protected by time - reversal , spatial rotation , and su(2 ) spin symmetries that coexist when @xmath34 . breaking the last of these symmetries by turning on spin - orbit coupling
thus produces a bulk gap given nearly by the _
atomic _ spin - orbit splitting for the adatoms , despite their dilute coverage .
this key point is demonstrated in fig .
[ bandstructurefig](d ) for @xmath35 ev , which yields a 0.32 ev gap that constitutes 80% of a single adatom
s spin - orbit splitting .
the gap opening indeed drives the system into a topological insulator phase .
since our hamiltonian is inversion symmetric this can be readily verified by computing fu and kane s formula for the @xmath36 invariant in ref .
@xcite . for additional evidence the solid curve in fig .
[ edgestatefig](a ) plots the density of states ( dos ) near zero energy for the same periodic adatom coverage on a graphene strip with armchair edges along @xmath37 and periodic boundary conditions along @xmath38 .
( our strip consists of 128 zig - zag ` rows ' of carbon sites , with 80 sites per row . )
edge states characteristic of the topological phase produce a finite dos inside of the bulk gap , and are clearly resolvable in the system size simulated over an energy window of 0.31 ev .
as an example , an edge state with mid - gap energy @xmath39 ev appears in fig .
[ edgestatefig](b ) , where circles indicate adatom locations while the shading represents the probability amplitude extracted from the wavefunction .
remarkably , the formation of a topological insulator by no means requires the periodic arrangements considered so far . in fact
similar physics arises even for _ completely randomly _ distributed h - site adatoms .
the dashed curve in fig .
[ edgestatefig](b ) illustrates the dos for the random case , again at 6.25% coverage . even in our finite system
one can easily resolve edge modes within a 0.21 ev energy range that is comparable to the bulk gap for the periodic case ; see , _
e.g. _ , the mid - gap state with energy @xmath40 ev plotted in fig .
[ edgestatefig](c ) .
next we demonstrate using density functional theory ( dft ) that the mechanism described above can be realized in graphene with 5@xmath0 adatoms , notably os and ir .
all dft calculations were carried out with the vienna ab - initio simulation package ( vasp ) @xcite at the level of the local density approximation ( lda ) @xcite , including spin - orbit coupling unless specified otherwise .
positions of all atoms were fully relaxed using the conjugated gradient method for energy minimization until the calculated force on each atom became smaller than 0.01 ev / .
most results were obtained using the supercell in fig .
[ bandstructurefig](a ) , with one adatom per cell . for additional details see .
@xmath41-grid mesh @xcite .
the energy cutoff of the plane wave expansion was set to 500 ev . ] below we report our results for os and then turn to ir .
since the thermal stability of adsorption structures is relevant for both experiments and applications , candidate adatoms should ideally exhibit large h - site binding energies [ defined as @xmath42 and high diffusion energy barriers [ defined as @xmath43 .
osmium satisfies both criteria .
the binding energy for os at the h site in graphene is 2.42 ev much larger than the ` top ' ( directly above a c ) and ` bridge ' ( above the midpoint of a c - c bond ) configurations for which @xmath44 is 1.70 and 1.59 ev , respectively .
moreover , the calculated diffusion barrier for an os adatom to diffuse from an h site through the top site is found to equal the difference between the binding energies at these positions , 0.72 ev .
the barrier for diffusion through the bridge site is similarly given by 0.83 ev .
therefore os adatoms should be stable over h sites even at room temperature .
by contrast most 3@xmath0 transition metals have @xmath45 ev @xcite and are much more mobile @xcite ; for example , the diffusion barrier is only 0.40 ev for co on graphene @xcite .
figure [ bands](a ) displays the dft band structure for periodic h - site os adatoms on graphene at 6.25% coverage using the @xmath24 supercell in fig .
[ bandstructurefig](a ) .
each os adatom forms a charge state of @xmath46 ( based on the bader charge division scheme @xcite ) , indicating that the os - graphene bonds mix covalent and ionic features .
clearly these bonds dramatically modify the characteristic dirac bands at the @xmath31 point of pure graphene similar to fig .
[ bandstructurefig ] .
most importantly , os induces a large band gap @xmath47 ev , _ right at the fermi level _ given by the green dashed line in fig . [
bands](a ) . as in our tight - binding model
the gap here results solely from spin - orbit coupling .
( without spin - orbit interactions a gapless spectrum arises ; see the supplemental material . ) more precisely , the partial density of states ( pdos ) for the os @xmath2 orbitals displayed in fig .
[ bands](b ) indicates that the gap arises from the hybridization between graphene s @xmath48 states and the spin - orbit - split @xmath4 and @xmath5 adatom orbitals , also as in our tight - binding model .
the pdos for the @xmath49 , @xmath50 , and @xmath51 orbitals , by contrast , is concentrated at much lower energies .
thus the gap - opening mechanism introduced earlier indeed appears in the realistic os / graphene system .
the os - induced gap depends exceptionally weakly on coverage .
to illustrate this important point we performed simulations of graphene with one os adatom in @xmath52 , @xmath53 , and @xmath54 supercells ( corresponding to coverages of 4% , 2.04% , and 1% ) .
circles in fig .
[ bands](c ) show the dft - predicted gaps , which remain close to 0.2 ev even at 1% coverage .
this striking feature is actually rather natural since the local atomic spin - orbit splitting for the os @xmath4 and @xmath5 orbitals essentially sets @xmath55 . strictly speaking
, a true topological insulator phase does not arise in the dft simulations described above since os forms small spin and orbital magnetic moments of 0.45 @xmath56 and 0.05 @xmath56 , respectively .
this produces visible splittings of the bands at the @xmath57 and m points corresponding to time - reversal - invariant momenta ; see fig .
[ bands](a ) .
one should keep in mind , however , that dft can sometimes overestimate moment formation .
nonetheless , even if a moment @xmath58 indeed appears in an experiment , there are practical means by which this can be quenched to zero to reveal a bona fide topological phase .
one effective approach is to apply an external electric field @xmath59 .
figure [ modifying](a ) illustrates that @xmath58 of os on graphene depends sensitively on @xmath59 . in particular @xmath60 transfers additional
charge from os to graphene and kills the moment for @xmath61 v / . the electric fields required to restore time - reversal symmetry only weakly affect the band structure .
see , for example , fig .
[ modifying](b ) corresponding to 6.25% os coverage with @xmath62 v / , where a time - reversal - invariant topological insulator appears with a gap @xmath63 ev .
co - doping provides another means to quench the os magnetic moment . to preserve the main features of the band structure while attracting charge away from os ( as accomplished by a negative @xmath59 ) , co - adsorbates should interact weakly with graphene and exhibit larger electronegativity than os . following this guidance , we considered cu , ag , and au in several configurations as described in the supplemental material .
whereas os repels ag and au adatoms , cu prefers to climb over os to form a vertical cu - os dimer over the h site .
the binding energy @xmath64 for these dimers is 5.96 ev , higher by 2.50 ev compared to that of well - separated cu and os adatoms .
additionally , cu more strongly anchors os to the h site since the binding energy for the vertical dimers over the top ( bridge ) site is weaker by 1.27 ( 1.42 ) ev .
this is not the only relevant criterion , however , since in practice isolated cu and os adatoms must be able to dimerize without overcoming substantial energy barriers .
we explored this issue by computing the total energies along the diffusion path depicted in fig .
[ modifying](c ) , where a cu adatom beginning at position d ends up above an os at position a. figure [ modifying](d ) illustrates the change in energy @xmath65 relative to the dimer state for various positions along this trajectory .
the energy barrier for a cu adatom to diffuse from location d to b is only @xmath66 ev ; once at position b the cu strongly attracts to the top of os and forms the vertical dimer.(d ) . ]
this suggests that dimer formation ought to proceed quite efficiently .
because of the hybridization and charge transfer between the cu and os atoms the bader charges of cu and os are respectively @xmath67 and @xmath68dft predicts that graphene with cu - os dimers is nonmagnetic . the spectrum for ( cu - os)/graphene at 6.25% coverage again supports a large topological insulator gap @xmath69 ev as evident in the band structure of fig .
[ modifying](e ) .
moreover , the triangles in fig . [ bands](c ) show that this gap exhibits similarly weak coverage dependence as for os / graphene .
the drawback here , however , is that the fermi level [ green line in fig .
[ modifying](e ) ] now resides in the valence band .
returning the fermi level to the insulating regime should be possible with conventional gating techniques , provided one works at low coverage .
alternatively , the hole introduced by each cu - os dimer can be compensated by replacing os with ir , which has one additional electron .
our calculations show that vertical cu - ir dimers also strongly bind to the h - site in graphene without forming a magnetic moment .
hybridization between cu - ir dimers and graphene produces nearly the same band structure as for ( cu - os)/graphene , but with the fermi level lying in the band gap .
see the band structure for 6.25% cu - ir coverage in fig .
[ modifying](f ) , where the gap is @xmath70 ev .
additional results for ir / graphene which behaves similarly to os / graphene can be found in the supplemental material . in summary ,
we have introduced a mechanism by which graphene covered with heavy adatoms realizes a topological insulator protected by a giant gap comparable to atomic spin - orbit energies , even at exceptionally dilute coverages .
using dft we predicted that os , ir , cu - os dimers , and cu - ir dimers all give rise to this mechanism and produce gaps exceeding 0.2 ev at coverages as low as 2% .
although our dft calculations of necessity invoked periodic adatom configurations , our tight - binding simulations indicate that readily observable bulk ( mobility ) gaps should survive also in the random case relevant for experiments .
these findings are expected to greatly facilitate the realization of a topological insulator phase in graphene - based systems .
we suspect , however , that the mechanism exposed here has much broader applications since ( contrary to ref .
@xcite ) the physics has nothing to do with the kane - mele model specific to graphene .
hybridizing trivial metals or insulators with heavy - element impurity bands may therefore provide a generic method for designing new topological phases , which would be interesting to investigate in future work . |
this work is supported by national science fund no 11374318 . | strict classic 3-d dynamics theory reveals that arbitrarily high center frequency light source can be achieved by flexible application of not - too - strong static electric field and static magnetic field .
the magnitudes of the fields are not required to be high . even though significant application value has promoted intensive investigation on short - wavelength light source ( or x - ray or higher energy photons source ) for decades [ 1 - 14 ] , there are still considerable difficulty in making a satisfactory achievement on this issue .
because @xmath0 @xmath1 ( @xmath2 ) wavelength radiation means the time cycle being @xmath3 ( @xmath4 ) , we therefore wish the time cycle of electron motion to be @xmath3 ( @xmath4 ) . clearly , the difficulty is just that so rapid electron oscillation , which can be finished hundreds ( or more ) times per @xmath5 , is unavailable .
this limits the efficiency of generating x - ray .
for example , in the high - order harmonics generation ( hhg ) , if the driving laser is of @xmath0 @xmath1 wavelength , the driven atomic dipole moment oscillation is usually of a time cycle @xmath6 .
the hhg is only because the time shape of the driven atomic dipole moment oscillation differs from @xmath7greatly .
this determines that low - order harmonics are dominant . according to most hhg experimental results [ 7 - 14 ] , low - order harmonics components whose orders are tens can be warranted but components at higher order hundreds are negligible . on the other hand , in free - electron - laser ( fel )
[ 2 - 5 ] , the efficiency of generating radiations at desired short wavelength is also far from satisfactory because sufficiently rapid electron velocity oscillation is unavailable .
clearly , if we can directly set up a sufficiently rapid electron velocity oscillation , we will obtain an efficient monocolor short - wavelength light source at desired short wavelength .
if @xmath8 wavelength is desired , we should set up an electron velocity oscillation whose time cycle is @xmath9 .
to drive so rapid oscillation usually demands , according to current knowledge , electromagnetic field of very short wavelength , which is unavailable because they are jsut what we are pursuing .
namely , we are trying to get what we are pursuing by applying what we are pursuing .
we consider a simple configuration containing merely static electric field ( along @xmath10-direction ) and static magnetic field ( along @xmath11-direction ) .
when an electron is input into this configuration , its behavior can be described by dimensionless 3-d relativistic newton equations ( rnes ) @xmath12 = 0,\ ] ] @xmath13 = w_bd_sx\ ] ] @xmath14 = -w_b\left [ \eta + d_sy\right]\ ] ] where @xmath15 moreover , @xmath16 and @xmath17 are constant - valued electric field and magnetic one and meet @xmath18 , @xmath19 and @xmath20 are reference wavelength and frequency , and @xmath21 @xmath22 @xmath23 @xmath24 @xmath25 is the cyclotron frequency , .
eqs.(1 - 3 ) will lead to @xmath26 @xmath27 & = & const = c_x , \end{aligned}\ ] ] where the values of these constants @xmath28 are determined from the initial conditions @xmath29 .
eqs.(5 - 7 ) can yield an equation set of @xmath30 and @xmath31 @xmath32 ^2*\left [ 1-\left ( d_sx\right ) ^2-\left ( d_sy\right ) ^2\right ] \\ \left ( d_sx\right ) ^2 & = & \left [ c_x - w_b*\left ( \eta s+y\right ) \right ] ^2*\left [ 1-\left ( d_sx\right ) ^2-\left ( d_sy\right ) ^2\right ] \end{aligned}\ ] ] whose solution reads @xmath33 ^2% } { \left [ 1+\left [ c_y+w_bx\right ] ^2+\left [ c_x - w_b*\left ( \eta s+y\right ) \right ] ^2\right ] } \ ] ] @xmath34 ^2}{\left [ 1+\left [ c_y+w_bx\right ] ^2+\left [ c_x - w_b*\left ( \eta s+y\right ) \right ] ^2\right ] } .\ ] ] note that the solutions ( 10,11 ) will cause @xmath35 ^2+\left [ c_x - w_b*\left ( \eta s+y\right ) \right ] ^2 } $ ] and , with the help of eqs.(6,7 ) , @xmath36 ( i.e. @xmath37 ) . if noting @xmath38 can be formally expressed as @xmath39 , which agrees with takeuchi s theory [ 15 ] , we can find that the electronic trajectory can be expressed as @xmath40 ^2=1+\left [ c_y+w_bx\right ] ^2+\left [ c_x - w_b*\left ( \eta s+y\right ) \right ] ^2,\ ] ] or @xmath41 ^2+\left [ \left ( y+\eta s\right ) -\upsilon _ { x0}\frac{\gamma _ 0}{w_b}\right ] ^2=\frac{\left [ \left ( \eta + \upsilon _ { y0}\right ) ^2+\left ( 1-\eta ^2\right ) \upsilon _ { x0}^2\right ] } { % 1-\eta ^2}\left ( \frac{\gamma _ 0}{w_b}\right ) ^2,\ ] ] where @xmath42 , @xmath43 and @xmath44
. there will be an elliptical trajectory for @xmath45 and a hyperbolic one for @xmath46 [ 15,16 ] .
the time for an electron travelling through an elliptical trajectory can be exactly calculated by re - writing eq.(10 ) as [ 15 ] @xmath47 where @xmath48 , @xmath49 \frac 1{w_b},$ ] @xmath50 and @xmath51 .
the equation can be written as a more general form @xmath52 where @xmath53 , @xmath54 and @xmath55 .
in addition , @xmath56 and @xmath57 .
it is easy to verify that for @xmath58 , there is @xmath59 .
initially , @xmath60 and hence @xmath61 . from strict solution
@xmath62 we can find the time for an electron travelling through an elliptical trajectory to meet @xmath63 $ ] and hence a time cycle @xmath64 namely , the oscillation along the elliptical trajectory will have a circular frequency @xmath65 .
moreover , it is interest to note that @xmath66 will lead to @xmath67 } { 1-\eta ^2}=0 $ ] and hence a straight - line trajectory @xmath68 .
the motion on an ellipical trajectory is very inhomogeneous .
the time for finishing the @xmath69 half might be very short while that for the @xmath70 half might be very long .
we term two halves as fast - half and slow - half respectively .
if @xmath71 is fixed over whole space , a fast - half is always linked with a slow - half and hence makes the time cycle for finishing the whole trajectory being at considerable level . for convenience ,
our discussion is based on the parameterized ellipse .
for the case @xmath72 , ( where @xmath73 is small - valued positive ) , the starting position @xmath74 is the left extreme of the ellipse and hence corresponds to @xmath75 . the time required for an acute - angled rotation from @xmath75 to @xmath76 , ( where @xmath77 is small - valued positive ) , will be @xmath78 + \sigma \sqrt{2\xi -\xi ^2}$ ] , which is @xmath79 if @xmath80 .
it is interest to note that if there is @xmath81 at the region @xmath82 , the electron will enter from @xmath83-region into @xmath84-region with an initial velocity whose @xmath10-component is @xmath85 and @xmath86-component @xmath87 is @xmath88 .
then , the electron will enter into the @xmath89-region a distance because of @xmath90 .
after a time @xmath91 , the electron will return into the @xmath83-region and the returning velocity will have a @xmath10-component @xmath92 . during this stage , the electron will move @xmath93 along the @xmath86-direction . then
, the motion in the @xmath94-region can be described by an acute - angled rotation along the ellipse @xmath95 .
thus , a complete closed cycle along the @xmath10-direction is finished even though the motion along the @xmath86-direction is not closed .
repeating this closed cycle will lead to an oscillation along the @xmath10-direction . clearly , the time cycle of such an oscillation is @xmath96 + 2\sigma \sqrt{% 2\xi -\xi ^2}$ ] . under fixed values of @xmath73 , @xmath97 and @xmath98 , the smaller
@xmath77 is , the smaller @xmath99 is .
there will be @xmath100 at @xmath80 . in principle
, arbitrary value of @xmath101 can be achieved by choosing suitable value of @xmath102 .
namely , arbitrary high center frequency ( @xmath103 ) oscillation can be achieved by choosing suitable value of @xmath102 .
although the time history of @xmath104 might cause its fourier spectrum being of some spread , the center frequency will be @xmath105 .
this result implies a simple and universal method of setting up quasi - monocolor light source at any desirable center wavelength .
that is , applying vertically static electric field @xmath106 and static magnetic field @xmath107 and on purpose letting a @xmath81 region existing and the ratio @xmath108 , then injecting electron along the @xmath86-axis with a velocity slightly above @xmath109 , and close to the boundary line between the @xmath110 region and the @xmath111 region .
as shown in fig.1 , adjusting the distance @xmath112 can lead to a quasi - monocolor oscillation source with any desired center frequency up to @xmath113-ray level .
in conclusion , we have described a simple and universal method of achieving quasi - monocolor light source at any desirable center wavelength .
the kernel of this method is to utilize the motion of fast electron near the surface of magnetic field .
here , fast electron means its velocity being @xmath114 .
actually , it is to adopt the fast - half of an ellipse and replace the time - consuming slow - half with a faster orbit governed by @xmath97 only . |
the nature of the interlayer coupling between two adjacent cuo@xmath4 planes in the cuprates is an important issue that remains unresolved .
suggestions include effects of strong intralayer scattering , @xcite non - fermi liquid ground states,@xcite the general phenomenon of confinement,@xcite inter- and in - plane charge fluctuations,@xcite indirect @xmath0-axis coupling through the particle - particle channel , @xcite as well as resonant tunneling on localized states in the blocking layer@xcite and two band models.@xcite coherent coupling originates from an overlap of the electronic wave functions between planes , and in - plane momentum is conserved in interlayer hopping .
by contrast for impurity - mediated incoherent coupling , the in - plane momentum is not constraint .
it has been shown that the @xmath0-axis conductivity sum rule depends on the nature of interlayer coupling.@xcite coherent coupling obeys the conventional sum rule@xcite regardless of the angular dependence of the interlayer hopping amplitude .
on the other hand , incoherent coupling violates the sum rule even if the in - plane dynamics can be described by a fermi liquid .
however , in order to explain the violation observed in some experiments,@xcite it was also necessary to include the non - fermi liquid nature of the in - plane dynamics.@xcite for yba@xmath4cu@xmath5o@xmath6 at optimum doping , a conventional @xmath0-axis sum rule is observed@xcite which is consistent with coherent @xmath0-axis coupling and an in - plane fermi liquid . for the underdoped case , a pseudogap is observed and the sum rule is closer to @xmath7.@xcite this value can most easily be understood as a pseudogap effect with incoherent @xmath0-axis coupling.@xcite another interesting example of the interlayer coupling is the @xmath0-axis conductivity due to low lying quasiparticles at zero temperature and zero frequency .
a method for observing the quasiparticle current is to measure the hysteretic @xmath0-axis tunneling current - voltage characteristics of layered cuprates.@xcite for measurements of the optical conductivity which includes the pair tunneling , see ref.@xcite for coherent coupling with a constant hopping amplitude @xmath1 , the @xmath0-axis quasiparticle conductivity @xmath8 shows a residual value independent of the in - plane scattering rate @xmath9 ( gap ) to leading order @xcite with the correction @xmath10 .
the underlying physics in this case is the same as for the in - plane quasiparticle conductivity;@xcite namely , the impurity - induced density of states at @xmath11 is canceled by the decrease of quasiparticle lifetime .
a universal value has been observed in yba@xmath4cu@xmath5o@xmath12 for the in - plane thermal conductivity.@xcite in some sense the thermal conductivity is an ideal probe of universal behavior since , as opposed to the electrical conductivity , it is not renormalized by vertex and by fermi - liquid corrections,@xcite and so there is less ambiguity in its identification .
the universal conductivity limit does not appear for coherent coupling with an angle - dependent hopping amplitude of the form @xmath13 which is believed to be appropriate for the copper oxides , where @xmath14 is an angle in the momentum space . in this case
the @xmath0-axis quasiparticle conductivity is reduced by a factor of @xmath15 as compared with the value of @xmath16 for a constant @xmath1 assuming the same magnitude of hopping amplitude , which of course is not the case .
this arises because the angular dependence , @xmath17 , eliminates the contribution of the quasiparticles on the nodal lines from interlayer transport . for incoherent coupling ,
the residual conductivity is proportional to @xmath18/\delta_{0 } \bigr)^{2}$ ] in leading order .
consequently , the universal value of the @xmath0-axis quasiparticle conductivity is a characteristic only of coherent coupling with a constant hopping amplitude .
geometrical consideration of the cu and o atomic arrangements@xcite from plane to plane leads one to expect that the dominant overlapping of orbitals would lead to a @xmath3 form with possibly a subdominant constant piece @xmath1 . in this case
, @xmath1 would still dominate the value of @xmath2 for very small @xmath19 , but the other terms would become important as @xmath20 corrections become larger in which case the term @xmath3 or the incoherent part can become important . in this paper
we investigate effects of interlayer couplings on the @xmath0-axis quasiparticle transport in the absence , as well as in the presence of an in - plane magnetic field .
it is our aim in this paper to look at all three contributions of interlayer coupling in detail , and our discussion of the magnetic field effects will not be restricted to the constant @xmath1 as it is in ref.@xcite we also investigate the role played by the conductivity sum rule in quasiparticle transport .
we find that it can be used to fix the relative amount of coherent to incoherent coupling .
we also find that @xmath21 shows a different @xmath22 coefficient for @xmath23 and @xmath24 a different @xmath25 coefficient @xmath26 depending on the nature of the coupling .
in the presence of an in - plane magnetic field , @xmath27 for coherent coupling with a constant hopping amplitude @xmath1 depends on the direction of the field@xcite as does the coefficient of the @xmath25 term in @xmath28 , but such a dependence is negligible for an angle - dependent coherent @xmath3 because it removes the contribution from the nodal quasiparticles which otherwise manifest the effect of field direction . for incoherent coupling , @xmath27 is independent of field direction , and in fact changes only when a paramagnetic interaction is included .
here we mention that there exist , in the literature , studies which consider effects of oxigen doping on the @xmath0-axis transport and coulomb charging effect@xcite associated with a pseudogap behavior.@xcite such complication , however , is beyond the scope of the present paper .
this paper is organized as follows : in sec .
ii , we derive general formulas associated with the @xmath0-axis quasiparticle conductivity including paramagnetic effects .
we describe , in sec .
iii , that effects of coherent coupling with a constant as well as with an angle - dependent hopping amplitude on the @xmath0-axis quasiparticle transport with and without an in - plane magnetic field . in sec .
iv , impurity - mediated incoherent @xmath0-axis coupling is considered .
we also illustrate the role of the conductivity sum rule in quasiparticle transport in sec . v and
draw conclusions in sec .
the hamiltonian @xmath29 for a cuprate superconductor with interlayer coupling can be written as @xmath30 , where @xmath31 describes a @xmath32-wave superconductor and @xmath33\;.\ ] ] the interlayer hopping amplitude @xmath34 depends on nature of @xmath0-axis couplings : i ) @xmath35 for coherent coupling with a constant amplitude .
ii ) for an angluar dependent amplitude @xmath36 . in the lattice
, it can be seen from geometrical consideration@xcite that @xmath37^{2}$ ] with an in - plane lattice constant @xmath38 .
ii ) for incoherent coupling @xmath39 and an impurity average is to be taken into account .
applying perturbation theory for @xmath40 , we obtain the @xmath0-axis quasiparticle conductivity @xmath41 in terms of the electronic spectral weight function @xmath42 .
@xmath43 a({\bf k},\epsilon)a({\bf p},\epsilon+\omega)\;,\ ] ] where @xmath44 is a constant which depends on the nature of interlayer coupling , @xmath45 is the usual fermi thermal factor , and the spectral function @xmath46 with @xmath47 .
we assume @xmath48 $ ] ( or @xmath49 in the continuum limit ) and a cylindrical fermi surface with @xmath50 measured from the fermi energy @xmath51 .
as @xmath52 and @xmath53 , @xmath54 .
it is easy to calculate @xmath55 for the couplings we mentioned earlier . to obtain the @xmath0-axis quasiparticle conductivity at finite frequency @xmath56 at zero temperature
, we take @xmath52 limit and assume @xmath23 @xmath57 a({\bf k},\epsilon)a({\bf p},\epsilon+\omega ) \nonumber\\ \simeq&&c\sum_{{\bf
k},{\bf p}}t^{2}_{{\bf k}-{\bf p } } a({\bf k},0)a({\bf p},\omega ) -c\sum_{{\bf
k},{\bf p}}t^{2}_{{\bf k}-{\bf p } } \left({\omega\over2}\right)\partial_{\epsilon}a({\bf k},\epsilon ) a({\bf p},\epsilon+\omega)\big|_{\epsilon=0 } \nonumber\\ + & & \ ; c\sum_{{\bf k},{\bf p}}t^{2}_{{\bf k}-{\bf p } } \left({\omega^{2}\over6}\right)\partial^{2}_{\epsilon } a({\bf k},\epsilon)a({\bf p},\epsilon+\omega)\big|_{\epsilon=0}+\cdots \label{finiteo}\end{aligned}\ ] ] where an expansion in @xmath56 has been made .
the finite @xmath58 corrections at @xmath11 will be considered later along with an in - plane magnetic field . in the presence of an in - plane magnetic field
, we assume that the field penetrates freely into the sample so that the field is uniform between the planes.@xcite this assumption should be valid for bi- and tl - based cuprates which are nearly two dimensional .
for ybco , however , there is a possibility that in - plane vortices form .
this means that in such a case an average over vortices in a unit cell is required.@xcite the interlayer coupling hamiltonian is modified by the presence of a uniform in - plane magnetic field @xmath59 as follows : @xmath60\;,\ ] ] where @xmath61 with an interlayer distance @xmath32 .
following a procedure similar to the one we applied in the zero field case , we obtain the @xmath0-axis quasiparticle conductivity @xmath62 in the presence of an in - plane magnetic field , @xmath63 a({\bf k}+{\bf q},\epsilon)a({\bf p},\epsilon+\omega)\;.\ ] ] in our consideration , the energy scale associated with the magnetic field are always much less than the gap @xmath64 . due to
an in - plane field , the quasiparticles gain an additional momentum @xmath65 on transferring to the next plane .
in other words , the bogoliubov - de gennes ( bdg ) wave functions@xcite @xmath66 and @xmath67 become @xmath68 and @xmath69 , respectively.@xcite the spectral weight function becomes @xmath70 where @xmath71 with @xmath72 .
note that the angular dependence of @xmath73 comes from the factor @xmath74 , where @xmath75 and @xmath76 is the direction of @xmath65 , which can be interpreted as the direction of the field @xmath77 because of the symmetry in the problem .
the gap seen by quasiparticles with a momentum @xmath78 is not @xmath79 but @xmath80 ; however , @xmath81 , where @xmath82 with @xmath83 , and the correction @xmath84 makes a negligible contribution to the quasiparticle energy spectrum . when the paramagnetic interaction is included , @xmath85 in the denominator of @xmath86 , where @xmath87 is the bohr magneton . as @xmath53 and @xmath52 , @xmath88 based on symmetry consideration,@xcite
one can easily deduce that the dependence of @xmath89 on field angle @xmath90 has a period @xmath91 because of the @xmath32-wave gap .
the sign of the gap does not matter .
consequently , @xmath92 where the @xmath93 s depend on the ratio of the magnetic field energy to @xmath19 _ i.e. _ , on @xmath94 and @xmath95 . since both @xmath96 and @xmath97 are linear to @xmath29 , we can parameterize the ratio @xmath98 .
we choose @xmath99 for the high-@xmath100 bi- and tl - based cuprates.@xcite another relation can be deduced without detailed calculation .
if we expand @xmath101 in terms of @xmath94 , then its angular dependence will appear for the first time from the term @xmath102 because the fourth harmonic @xmath103 comes from the @xmath104 term .
this means that @xmath105 is required for any clear angular dependence to show up . for the eighth harmonic @xmath106 to enter
, the field has to be even higher . at a finite temperature @xmath107
, we apply the sommerfeld expansion@xcite to @xmath108 and obtain @xmath109\ ; , \label{finitet}\end{aligned}\ ] ] where @xmath110 with @xmath111 .
we use the nodal approximation@xcite to describe the low temperature physics associated with the quasiparticle transport because quasiparticles near the nodal lines dominantly contribute to the resistive transport .
linearizing @xmath112 and @xmath79 near the nodal points on the fermi surface ( fs ) , we obtain @xmath113 and @xmath114 , where @xmath115 is the fermi velocity ( momentum ) and @xmath116 is the gap velocity .
@xcite the unit vector @xmath117 ( @xmath118 ) is perpendicular ( parallel ) to the fs .
these unit vectors will change depending on the nodal line .
for example , on the nodal line of @xmath119 , @xmath120 and @xmath121 . since we include quasiparticles only near the nodal points on the fs in our considerations , we will use the following procedure in the actual calculation .
@xmath122 where @xmath123 , @xmath124 and @xmath125^{-1}$ ] .
the coordinate transformations we made are rotation , translation and dilation . in the coordinate of @xmath126 ,
the energy dispersion of the quasiparticle @xmath127 becomes @xmath128 , where @xmath129 .
in this section , we consider the effects of coherent coupling on @xmath0-axis quasiparticle transport . since @xmath35 for a constant hopping amplitude , as @xmath53 and @xmath52 , @xmath130 . for @xmath23 , we use the nodal approximation for eq .
( [ finiteo ] ) . then , up to order @xmath131 , @xmath132 where @xmath133 is a constant independent of @xmath19 . in the presence of a uniform in - plane field
, the quasiparticle conductivity becomes @xmath134 applying the substitution eq .
( [ nodal ] ) to the above equation , we obtain without the paramagnetic part @xmath135 it can be shown within the nodal approximation that @xmath136 $ ] , where @xmath137 is a direction on the nodal line ; however , to a good approximation , we can replace @xmath138 by @xmath137 . for the weak field case @xmath139 ,
we expand @xmath101 in terms of @xmath94 and obtain @xmath140\left({e_{q}\over\gamma}\right)^{4}\;,\ ] ] where @xmath133 is the @xmath0-axis quasiparticle conductivity in the absence of the in - plane magnetic field . as we see , the angular dependence of @xmath101 is small for a weak field .
@xmath101 is maximum ( minmum ) when the in - plane field is along a nodal ( anti - nodal ) line .
it can be physically understood as follows : when * h * is along a nodal line , for example , @xmath141 , the angle of @xmath65 is @xmath142 because @xmath143 .
then , while the shifted momenta @xmath78 of the quasiparticles with @xmath119 and @xmath144 deviate from the nodal regions when @xmath145 is compatible with @xmath146 , quasiparticles with @xmath147 and @xmath148 remain in the nodal areas .
this means the remaining quasiparticles govern the @xmath0-axis transport when @xmath141 . on the other hand , for @xmath149 , all quasiparticles move away from the nodal regions due to the momentum shift @xmath65
therefore , @xmath150 and @xmath151 because of a mismatch between @xmath152 and @xmath78 in the interlayer transport . in fig . 1 , we plot resistivity as a function of direction of the in - plane magnetic field , @xmath153 for various values of @xmath96 with ( solid line ) and without ( dashed line ) the paramagnetic interaction included
. we reproduced results of ref.@xcite for @xmath154 , @xmath155 , and @xmath156 .
as shown in fig . 1 , @xmath157 is increased as the magnetic field increases , and it is decreased for a give @xmath96 when the paramagnetic interaction is considered because the interaction is pair breaking . if we consider only the paramagnetic interaction , then the quasiparticle conductivity @xmath158 has no @xmath90 dependence and increases as follows : @xmath159\;.\ ] ] the eighth harmonic clearly appears for @xmath160 .
actually , it begins to appear when @xmath161 .
interestingly , the paramagnetic interaction unambiguously reduces the amplitude of @xmath157 for a higher field @xmath162 .
we also plot @xmath163 of eq .
( [ finitet ] ) for a finite @xmath107 in fig . 2 .
for reference in the absence of the magnetic field , @xmath164 ( dashed line ) . in the case of an angle - dependent hopping amplitude @xmath165 ,
the @xmath0-axis quasiparticle conductivity in zero field is @xmath166 . for @xmath23 , using the nodal approximation we obtain @xmath167 where @xmath168 in this case .
effects of the in - plane field on the @xmath0-axis quasiparticle conductivity can be seen in the same way as before , namely , @xmath169 since @xmath170 within the nodal approximation , @xmath101 becomes without the paramagnetic interaction @xmath171 in this case , we find that @xmath172 without clear dependence on the direction of the in - plane field , and @xmath173 $ ] in the leading order and the next order is @xmath174 $ ] ; therefore , @xmath163 is negligible .
the @xmath0-axis quasiparticle conductivity is insensitive to the in - plane magnetic field because the angle - dependent hopping amplitude @xmath175 eliminates contributions of nodal quasiparticles , which otherwise manifest the effects of the field .
for impurity - mediated incoherent coupling @xmath176 , we need a model for the impurity scattering potential and need to carry out an average over impurity configurations . we use a simple model@xcite for the scattering potential @xmath177 with respect to scattering symmetry so that it is decomposed @xmath178\cos[(4l'+2)\phi_{p}]\;,\ ] ] where @xmath179 and @xmath180 are integers
. however , for the @xmath0-axis quasiparticle transport , only the @xmath181 term gives a non - zero value to the conductivity .
as @xmath53 and @xmath52 , @xmath182 . for @xmath23 , it can be shown that @xmath183 where @xmath184^{2}$ ] for impurity - mediated incoherent coupling . in the presence of an in - plane magnetic field
, the @xmath0-axis quasiparticle conductivity becomes @xmath185 as we see , the angular dependence of @xmath101 is determined by @xmath186 .
it can be shown numerically and analytically that the field effect appears only as the paramagnetic contribution .
we present some analytic results , for example , in the weak field limit @xmath187\simeq { 2\gamma\over{\pi v_{g}v_{f}}}\;h\left({\mu_{b}h\over\gamma}\right)\;,\ ] ] where @xmath188 .
therefore , the @xmath0-axis quasiparticle conductivity for a weak field becomes @xmath189 consequently , for incoherent coupling the in - plane magnetic field has no effect on the @xmath0-axis quasiparticle transport if the paramagnetic interaction is not considered .
physically , the momentum of the transport quasiparticle is not constraint in incoherent coupling so that the change in momentum @xmath65 does not matter . since the paramagnetic interaction is pair breaking ,
the magnitude of @xmath101 is increased but shows no angular dependence when this interaction is included . at a finite temperature @xmath26 ,
we use eq .
( [ finitet ] ) and for a weak field , @xmath163 becomes @xmath190 \left[1-(\mu_{b}h/\gamma)^{2}\right]}}\;.\ ] ]
from the @xmath0-axis conductivity sum rule , the superfluid density @xmath191 can be written in terms of the missing spectral weight @xmath192 and the theral averages of kinetic energy @xmath193 of a superconducting ( normal ) state as follows : @xmath194,\ ] ] where @xmath195 is the cutoff frequency for the interband transitions that @xmath40 does not describe . using the above equation as well as the kramers - kronig relation between the conductivity and the penetration depth
, we obtain the normalized missing spectral weight @xmath196 ; @xmath197\over \sum_{\omega}\sum_{{\bf k},{\bf p } } where @xmath198 and @xmath199 are superconducting green functions and @xmath200 is in the normal state .
@xcite since , in principle , all possible interlayer couplings might be present in a given sample , we need to consider coherent and incoherent coupling at the same time .
the actual @xmath0-axis transport of quasiparticles is determined from competing effects between couplings .
based on the conductivity sum rule , we can estimate the ratio @xmath201 , where @xmath202 @xmath203 is the @xmath0-axis quasiparticle conductivity for coherent ( incoherent ) coupling .
since the sum rule does not distinguish between @xmath1 and @xmath3 , we consider two limiting cases : i ) @xmath204 and ii ) @xmath205 . as @xmath53 and @xmath52 , for the case with @xmath1 @xmath206 \nonumber\\ = & & \sigma_{co}\left[1+{4\delta_{0}\sigma_{cn}\over \pi e^{2}dt^{2}_{\perp}n(0)}\left({\gamma\over\delta_{0 } } \ln(\delta_{0}/\gamma)\right)^{2}\right]\;,\end{aligned}\ ] ] where we have used @xmath207 and @xmath208^{2}$ ] with @xmath209 .
one may think that we need to know the ratio of @xmath181 to @xmath210 in order to compare @xmath202 with @xmath211 .
however , the conductivity sum rule helps us to estimate the ratio @xmath201 without _ ad hoc _ information .
let us define @xmath212 , where @xmath191 and @xmath213 are the corresponding superfluid density and penetration depth , respectively . for coherent coupling @xmath214\;,\ ] ] where @xmath215 is the complete elliptic integral of the first kind . for incoherent coupling
, we assume @xmath216 for simplicity and to illustrate possible effects .
then , @xmath217^{2}\;,\ ] ] where @xmath218 , @xmath219 , and @xmath220 is the complete elliptic integral of the second kind.@xcite since @xmath221 , @xmath222 $ ] and we obtain @xmath223\;. \label{cond1}\ ] ] since the total superfluid density @xmath224 , it can be shown that @xmath225 note that @xmath226 in eq .
( [ cond1 ] ) can be determined from the violation of the conductivity sum rule eq .
( [ sum ] ) .
@xmath227 corresponds to pure coherent @xmath0-axis coupling in which case the sum rule eq .
( [ sum ] ) is conventional and equal to @xmath228 . on the other hand ,
@xmath229 means pure incoherent coupling and the sum rule is larger than @xmath228 and equal to @xmath230 in our model with @xmath231 .
a finite @xmath226 corresponds to an admixture of coherent and incoherent @xmath0-axis transport .
for example , @xmath232 applies when @xmath233 , that is , there is a @xmath234 violation of the sum rule upwards .
the sum rule itself determines the admixture of @xmath235 to @xmath236 , but can not differentiate between @xmath1 and @xmath3 for the coherent part .
limits on the relative size of these two overlap integrals can only be set from consideration of the chemistry of the cuo@xmath4 planes and their overlap or , alternatively , from experimental information such as the behavior of @xmath2 when the impurity content is increased .
for the case with @xmath3 @xmath237\;,\ ] ] where @xmath220 is the complete elliptic integral of the second kind , and @xmath226 as above with @xmath238 and @xmath239 . following a similar way as before ,
we obtain @xmath240\ ; , \label{cond2}\ ] ] where @xmath241 with @xmath242 and @xmath239 .
note that @xmath1 and @xmath243 case both contribute in the same order to the zero temperature @xmath0-axis superfluid density . for the pure case ,
assuming @xmath1 and @xmath243 are of the same magnitude , the ratio of the two contributions is @xmath244 .
this is in sharp contrast to what we found in the case of @xmath2 where @xmath1 contributes of order @xmath210 while the contribution from @xmath243 is of order @xmath245 and so vanishes in the clean limit .
recently , gaifullin _
et al._@xcite have measured the temperature dependence of the @xmath0-axis superfluid density in bi@xmath246 and found that at low temperature it is well fit by a form @xmath247 $ ] with @xmath248 of order @xmath249 close to the values reported for hg@xmath250.@xcite this finding favours a pure @xmath3 model which is known to give a @xmath251 law ( ref.@xcite ) and leaves little room for a subdominant @xmath1 contribution because in this case the @xmath0-axis penetration depth will mirror perfectly@xcite its in - plane temperature dependence which goes like @xmath58 .
the data certainly preclude a linear @xmath58 contribution to the superfluid density of more than a few percent implying a ratio @xmath252 of order @xmath253 at most . in this instance
, the ratio @xmath254 can easily be comparable to @xmath252 and can even be the larger of the two . as an illustration in the experimental work establishing the in - plane universal limit for the thermal conductivity@xcite the values of @xmath254 are of order @xmath255 .
latyshev _ et al._@xcite quoted @xmath256mev for their single crystals , which is of similar order .
this implies that in the experiments on bi@xmath246 , the contribution to @xmath2 of each of @xmath1 and @xmath243 terms are likely to be close in magnitude . nevertheless , it is concluded in ref.@xcite that the universal limit is observed so that @xmath1 presumably still dominates .
additional work with various values of @xmath19 , somewhat larger as well as smaller than used so far , should allow one to establish the size of the important ratio @xmath252 .
this information , with the @xmath0-axis optical sum rule , would allow an estimate of each of the three possible contributions to the @xmath0-axis transport discussed in this paper .
these parameters are important because they control all @xmath0-axis transport .
we point out here that we have considered separately the case @xmath1 and @xmath3 .
if amplitudes are added and squared , a cross term of the form @xmath257 will enter but this will not change qualitatively the conclusion made above .
the contribution of such a term to @xmath133 is @xmath258 and to the superfluid density is @xmath259 $ ] .
we have discussed how the nature of the interlayer coupling influences the @xmath0-axis quasiparticle conductivity @xmath62 in the absence and in the presence of an in - plane magnetic field . in zero field
, @xmath2 is independent of the in - plane scattering rate @xmath19 , to leading order , only for coherent coupling with a constant hopping amplitude .
also , @xmath260 shows a different behavior for different @xmath0-axis coupling .
similarly , the field effects depend crucially on the nature of the interalyer coupling . for coherent coupling with a constant hopping amplitude ,
an angular dependence of the conductivity @xmath101 appears in high field .
the resistivity @xmath157 increases because the mismatch between @xmath152 and @xmath78 decreases @xmath101 , and @xmath157 is maximum ( minimum ) when the field is along the anti - nodal ( nodal ) line .
this confirms previous work.@xcite when the paramagnetic interaction is included , @xmath157 is decreased because this interaction is pair breaking and the anisotropy also decreases . for high field @xmath162 , the eighth harmonic appears and can be seen in the @xmath90 dependence of @xmath157 . for coherent coupling with an angle - dependent hopping amplitude ,
@xmath101 is insensitive to field direction because this hopping amplitute eliminates the contribution of the nodal quasiparticles which manifest the angular dependence .
also , the paramagnetic interaction does not significantly change @xmath101 .
for incoherent coupling , only the paramagnetic interaction has an effect on the @xmath0-axis quasiparticle transport and @xmath101 has no dependence on field direction .
the coefficient of the first temperature correction which goes like @xmath25 is anisotropic only in the constant @xmath1 case and shows the same minima ( maxima ) as does the leading contribution .
the @xmath0-axis conductivity sum rule helps estimate separately the contributions from coherent and incoherent coupling to the quasiparticle transport without microscopic information on hopping amplitudes .
however , it can not , by itself , differentiate between contributions from constant or angular dependent coherent hopping amplitude . to get separate information on these two contributions ,
impurity studies could be used .
consideration of the temperature dependence of the @xmath0-axis superfluid density measured in josephson plasma resonance experiments already gives evidence that @xmath243 is much larger than @xmath1 , but do not unambiguously rule out a small @xmath1 contribution which could still be large enough to dominate the universal limit for @xmath261 in relatively pure samples @xmath262 . at the same time
, @xmath243 would dominate the temperature dependence of the superfluid density .
we mention that the sum rule of eq .
( [ sum ] ) was derived under the assumption that the normal state is fermi - liquid like . however
, if it has a non - fermi - liquid nature such as a pseudogap , the sum rule has to be modified to account for this ; therefore , it is necessary to explore the competing effects of interlayer coupling in a more fundamental theory which goes beyond a fermi liquid normal state .
research was supported in part by the natural science and engineering research council of canada ( nserc ) and by the canadian institute for advanced research ( ciar ) .
w. k. thanks e. h. kim for many useful discussions .
p. j. hirschfeld , w. o. putikka , and d. j. scalapino , phys .
b * 50 * , 10250 ( 1994 ) ; m. j. graf _ et al . _ ,
b * 52*,10588 ( 1995 ) ; _ ibid _ * 53 * 15147 ( 1996 ) ; w. c. wu , d. branch , and j. p. carbotte , phys . rev .
b * 58 * , 3417 ( 1998 ) ; i. schrrer , e. schachinger , and j. p. carbotte , j. low temp . phys . * 115 * , 251 ( 1999 ) .
this means the momentum of cooper pair is still zero . in order to include a finite momentum ( @xmath263 ) state of a cooper pair , bdg wave functions @xmath66 and
@xmath67 become @xmath68 and @xmath264 , respectively .
however it has been shown that such a state is unstable under the field much less than the upper critical field , which is the case we consider .
see yang and sondhi , phys .
b * 57 * , 8566 ( 1998 ) . | the @xmath0-axis quasiparticle conductivity shows different behavior depending on the nature of the interlayer coupling . for coherent coupling with a constant hopping amplitude @xmath1 , the conductivity at zero frequency and zero temperature @xmath2 depends on the direction of the magnetic field , but it does not for angle - dependent hopping @xmath3 which removes the contribution of the nodal quasiparticles . for incoherent coupling ,
the conductivity is also independent of field direction and changes only when paramagnetic effects are included .
the conductivity sum rule can be used to determine the admixture of coherent to incoherent coupling .
the value of @xmath2 can be dominated by @xmath1 while at the same time @xmath3 dominates the temperature dependence of the superfluid density . |
quasi - exactly solvable ( qes ) schr " odinger equations occupy an intermediate place between exactly - solvable and non - solvable ones . such equations , which have widespread applications in physics , chemistry , and engineering , are characterized by the fact that a ( finite ) part of their spectrum can be determined algebraically if the potential parameters satisfy certain constraints , while the remaining part remains unknown . among the procedures used to construct them ,
one may quote the recursion relation method , which is rather hard to perform since one has to deal with three- or more - step recursion relations ( see , e.g. , @xcite ) .
two easier approaches are the one based on an sl(2,@xmath0 ) algebraization @xcite and that relying on the functional bethe ansatz method @xcite .
the most prominent example of qes systems is the sextic polynomial oscillator @xcite .
in contrast , the non - qes nature of the simpler and phenomenologically more important quartic polynomial oscillator @xcite has always been surprising . in a recent work inspired by the well - known example of the square
well , znojil proposed a resolution of this paradox by allowing an analyticity - violating symmetrization of the quartic polynomial oscillator and he proved that the resulting potential is qes by using the recursion relation approach @xcite .
this has emphasized the interest of splitting the real line of coordinates into subintervals and considering potentials that admit different definitions inside these subintervals , while remaining continuous on the whole real line @xcite .
such a procedure indeed opens the possibility of constructing new es or qes potentials .
the purpose of the present work is twofold : first to re - examine the symmetrized quartic polynomial oscillator in the light of the sl(2,@xmath0 ) lie algebraic method , and second to propose a new symmetrized sextic one , while providing some explicit solutions of both of them .
as in @xcite , let us consider the symmetrized quartic polynomial oscillator @xmath1 which is non - analytic and defined separately on the two semi - infinite intervals @xmath2 and @xmath3 .
here @xmath4 , @xmath5 , and @xmath6 are three real constants . the corresponding schr " odinger equation @xmath7 may be considered separately on these two intervals .
physically acceptable wavefunctions should tend to zero at infinity and be continuous everywhere , as well as their first derivative .
since the potential is symmetric under parity , they must also be either symmetric or antisymmetric .
we may therefore solve eq .
( [ eq : se ] ) in one of the intervals , e.g. , @xmath2 , then impose that in the other interval @xmath3 , @xmath8 with @xmath9 or @xmath10 according to whether @xmath11 is even or odd .
qes solutions of ( [ eq : se ] ) may be obtained by assuming that @xmath12 takes the form @xmath13 with @xmath14 here @xmath15 , @xmath16 are two real constants , and @xmath17 ( @xmath18 , 1 , , @xmath19 ) are some expansion coefficients to be determined .
such functions have the right behaviour for @xmath20 . on the other hand
, the continuity of @xmath21 and of its derivative at the origin imposes the additional condition @xmath22 inserting eq .
( [ eq : qes-4 ] ) in ( [ eq : se ] ) and setting @xmath23 in order to eliminate some high - order terms , we arrive at the differential equation @xmath24 on the interval @xmath2 .
this is a ( triconfluent ) heun equation @xcite .
hence , provided we also set @xmath25 the second - order differential operator @xmath26 on its left - hand side can be written as a second - degree polynomial @xcite @xmath27 in the sl(2,@xmath0 ) generators @xmath28 acting in the ( @xmath29)-dimensional representation in the space of polynomials of order not higher than @xmath19 , @xmath30 .
on using the relations @xmath31 , @xmath32 , @xmath33 , it is straightforward to explicitly write the matrix of @xmath26 in the space @xmath34 .
its diagonalization , followed by the imposition of constraint ( [ eq : constraint-4 ] ) , will then provide the admissible energies @xmath35 and wavefunctions @xmath21 . in @xcite
, znojil solved the rather complicated @xmath36 , @xmath37 case by the recursion relation method . in the remainder of this section
, we plan to exhibit some simpler examples , which can be directly derived by using ( [ eq : heun])([eq : j ] ) . for @xmath38
, we obtain @xmath39 and @xmath40 on @xmath2 .
since @xmath41 and @xmath42 , according to ( [ eq : constraint-4 ] ) , the resulting wavefunction on the whole real line can not be odd , but it can be even if the condition @xmath43 is satisfied .
we conclude that the potential @xmath44 has an energy eigenvalue @xmath45 and associated wavefunction @xmath46 , given by @xmath47 corresponding to a ground state .
turning now ourselves to the @xmath48 case , we arrive at a second - degree equation for @xmath49 , whose solutions @xmath50 are real provided @xmath51 , which imposes @xmath52 or @xmath53 and either @xmath54 or @xmath55 .
the corresponding eigenfunctions on @xmath2 are @xmath56 . hence @xmath57 and @xmath58 . on the whole line
, we can get an odd wavefunction if @xmath59 , which imposes @xmath43 and @xmath60 ( resp .
@xmath61 ) for @xmath62 ( resp . @xmath63 ) .
we conclude that the potential @xmath64 has an energy eigenvalue @xmath65 and associated wavefunction @xmath66 , given by @xmath67 corresponding to a first - excited state
. we may also ask whether an even wavefunction can be obtained for @xmath48 . from ( [ eq : constraint-4 ] ) , we directly get that @xmath68 and @xmath69 , where the upper ( resp .
lower ) sign corresponds to @xmath62 ( resp .
@xmath63 ) .
this implies that @xmath15 can be expressed in terms of @xmath16 as @xmath70 . as a result , @xmath71 and @xmath72 if @xmath73 or @xmath74 , whereas @xmath75 and @xmath76 if @xmath77 . in both cases , @xmath78
hence , the potential @xmath79 has an energy eigenvalue @xmath80 and associated wavefunction @xmath81 , given by @xmath82 corresponding to a ground state for @xmath83 and to a second - excited state for @xmath73 .
, defined in eq . ( [ eq : v-0 + ] ) , in terms of @xmath84 for @xmath85 ( black line ) , @xmath86 ( red line ) , and @xmath87 ( blue line ) . in all cases , @xmath88 and @xmath89 . ] , defined in eq .
( [ eq : psi-0 + ] ) , in terms of @xmath84 for @xmath85 ( black line ) , @xmath86 ( red line ) , and @xmath87 ( blue line ) . in all cases , @xmath88 and @xmath89 . ] , defined in eq .
( [ eq : v-1 + ] ) , in terms of @xmath84 for @xmath86 , @xmath88 , @xmath90 ( red line ) , and @xmath87 , @xmath88 , @xmath91 ( blue line ) .
the corresponding energy is @xmath92 and @xmath93 , respectively . ] , defined in eq .
( [ eq : psi-1 + ] ) , in terms of @xmath84 for @xmath86 , @xmath88 , @xmath90 ( red line ) , and @xmath87 , @xmath88 , @xmath91 ( blue line ) .
the corresponding energy is @xmath92 and @xmath93 , respectively . ]
let us now consider the symmetrized sextic polynomial oscillator @xmath94 where @xmath4 , @xmath5 , @xmath6 , @xmath95 , and @xmath96 are five real constants , and let us proceed as in sec . 2 .
qes solutions of ( [ eq : se ] ) with potential ( [ eq : sextic ] ) may be obtained by assuming that @xmath97 where @xmath15 , @xmath16 , @xmath98 are three real constants and @xmath99 is still given by eq .
( [ eq : phi ] ) , but with eq .
( [ eq : constraint-4 ] ) replaced by the constraint @xmath100 by setting now @xmath101 we get on @xmath2 the differential equation @xmath102 which is ( a special case of ) a generalized heun equation .
its @xmath19th - degree polynomial solutions @xmath103 with distinct roots @xmath104 , @xmath105 , , @xmath106 , can be found by applying the functional bethe ansatz method .
direct application of theorem 1.1 of ref .
@xcite leads to the following conditions @xmath107 where the roots @xmath104 , @xmath105 , , @xmath106 satisfy the bethe ansatz equations @xmath108 on the resulting @xmath109 , it then remains to impose constraint ( [ eq : constraint-6 ] ) .
the @xmath38 is straightforwardly solved . as for the quartic oscillator
, there is no odd eigenfunction , but there is an even one provided @xmath110 .
hence , the potential @xmath111 has an energy eigenvalue @xmath45 and associated wavefunction @xmath46 , given by @xmath112 corresponding to a ground state . for @xmath48
, there is a single bethe ansatz equation to be solved , namely @xmath113 . since @xmath114 and @xmath115
, we can get an even or odd wavefucntion by appropriately choosing the parameters .
the function is odd if @xmath116 , implying @xmath110 .
hence the potential @xmath117 has an energy eigenvalue @xmath65 and associated wavefunction @xmath118 , given by @xmath119 corresponding to a first - excited state . on the other hand ,
the function is even if @xmath120 , where @xmath98 must be a nonvanishing solution of the quartic equation @xmath121 .
then , for such a @xmath98 value , the potential @xmath122 admits @xmath80 and @xmath81 , given by @xmath123 representing a ground state for @xmath124 and a second - excited state for @xmath125 . for @xmath36 and @xmath126 , we get @xmath116 and @xmath127 , where @xmath98 is a nonvanishing solution of @xmath128 .
this yields @xmath129 where @xmath130 corresponds to a first - excited ( resp .
third - excited ) state for @xmath124 ( resp .
@xmath125 ) . furthermore , for @xmath36 and @xmath37 , we obtain @xmath131 where
@xmath104 and @xmath105 are solutions different from zero and @xmath132 of the set of equations @xmath133 the set of qes symmetrized sextic polynomial oscillators that we have constructed here can be separated into two categories . *
the symmetrized sextic oscillators whose known wavefunction is of ` natural ' parity , i.e. , such that @xmath134 .
these include those associated with the cases @xmath135 , @xmath136 , and @xmath137 , considered above .
it is straightforward to see that they comprise as special cases the well - known ( analytic ) sextic oscillators of ref .
this is obvious if we set @xmath85 in eqs .
( [ eq : v-0+])([eq : e-1- ] ) , corresponding to @xmath135 and @xmath136 . in the case of eq .
( [ eq : v-2 + ] ) for @xmath138 , on assuming @xmath139 , the set of equations ( [ eq : bethe ] ) reduces to a single equation @xmath140 , leading to @xmath141 . as a consequence , @xmath142 and @xmath143 , in agreement with @xcite . * the symmetrized sextic oscillators whose known wavefunction is of ` unnatural ' parity , i.e. , such that @xmath144 .
these include those associated with the cases @xmath145 and @xmath146 , considered above .
such symmetrized oscillators do not comprise any analytic sextic oscillator because the @xmath147 condition always ensures the nonvanishing of an @xmath148 term in @xmath149 . in figs . 1 and 2 ,
some examples of potential and associated wavefunction are displayed for the case @xmath135 , belonging to the first category .
they are compared with the corresponding results for the analytic potential with @xmath85 . in figs . 3 and 4 ,
some examples of potential and associated wavefunction are plotted for the case @xmath145 , belonging to the second category .
in the present work , we have shown that the previously introduced qes symmetrized quartic polynomial oscillator admits an sl(2,@xmath0 ) algebraization and we have used the latter to provide some simple examples of bound - state wavefunctions at certain couplings and energies .
we have then proposed a new symmetrized sextic polynomial oscillator and we have proved its qes nature by deriving some exact , closed - form solutions of the corresponding schrdinger equation by resorting to the functional bethe ansatz method .
furthermore , we have shown that such symmetrized sextic oscillators can be separated into two categories : a first set generalizing the well - known analytic sextic oscillators , which form a subset , and a second set of novel qes potentials with no counterpart in the class of analytic sextic oscillators . | the symmetrized quartic polynomial oscillator is shown to admit an sl(2,@xmath0 ) algebraization . some simple quasi - exactly solvable ( qes ) solutions are exhibited .
a new symmetrized sextic polynomial oscillator is introduced and proved to be qes by explicitly deriving some exact , closed - form solutions by resorting to the functional bethe ansatz method .
such polynomial oscillators include two categories of qes potentials : the first one containing the well - known analytic sextic potentials as a subset , and the second one of novel potentials with no counterpart in such a class . plus 1pt minus 1pt running head : symmetrized polynomial oscillators keywords : quantum mechanics ; quasi - exact solvability ; polynomial oscillators pacs nos .
: 03.65.fd , 03.65.ge |
the russian spectrum - roentgen - gamma ( srg ) satellite will fly on a medium class platform ( navigator , lavochkin association , russia )
. the launch will be in 2012 using a soyuz-2 rocket from bajkonur into an orbit around l2 .
the payload consists of the x - ray instruments erosita ( extended roentgen survey with an imaging telescope array ) and art - xc ( astronomical roentgen telescope x - ray concentrator ) .
the seven erosita telescopes are based on the existing design launched on the abrixas mission plus an advanced version of the pnccd camera successfully flying on xmm - newton . in order to optimize erosita for the dark energy studies ,
the effective area is increased by a factor of five , the angular resolution is improved by a factor of two , and the field of view is also increased by a factor of two with respect to abrixas .
such a design has been drawn to match the outcome of the most recent calls for ideas on dark energy observations ( like e.g. by nasa , doe , esa , eso and others ) .
similarly to erosita art - xc contains 7 telescopes working in the energy range between 6 and 30 kev .
the telescopes are conical approximations of the wolter - i geometry with cdte detectors in their focal planes ( pavlinsky et al . , 2010 , this volume )
the erosita telescope on board srg . @xmath6 the grasp of erosita compared with rosat - pspc and xmm - newton.,title="fig : " ] the erosita telescope on board srg . @xmath6 the grasp of erosita compared with rosat - pspc and xmm - newton.,title="fig : " ] one way to test cosmological models and to assess the origin , geometry , and dynamics of the universe is through the study of the large - scale structures indeed galaxy clusters are strongly correlated and thus they are good tracers of the large - scale structure on very large scales by sampling the most massive congregates of matter .
the galaxy cluster population provides information on the cosmological parameters in several complementary ways : 1 .
the cluster mass function in the local universe mainly depends on the matter density @xmath7 and the amplitude of the primordial power spectrum @xmath8 .
2 . the evolution of the mass function f(m , z )
is directly determined by the growth of structure in the universe and therefore gives at the same time sensitive constraints on dark matter and dark energy .
3 . the amplitude and shape of the cluster power spectrum , p(k ) and its growth with time , depend sensitively on dark matter and dark energy .
baryonic wiggles due to the acoustic oscillations at the time of recombination are still imprinted on the large scale distribution of clusters ( i.e. in their p(k ) and the autocorrelation function ) and thus can give tight constraints on the curvature of space at different epochs .
the constraints provided by the different cosmological tests with clusters are complementary in such a way , that degeneracies in the parameter constraints in any of the tests can be broken by combinations .
the simultaneous constraint of @xmath7 and @xmath8 by combining method 1 and 3 above is one such example @xcite .
in addition the combination of several tests provides important consistency checks as explained below .
in addition to the above applications , galaxy clusters have been used as cosmological standard candles to probe absolute distances , analogous to the cosmological tests with supernovae type ia .
the assumption that the cluster baryon fraction is constant with time combined with observations of this quantity provides constraints on dark matter and dark energy , e.g. @xcite ) . in a very similar way , combined
x - ray and sunyaev - zeldovich - measurements provide a mean for absolute distance measurements and constraints of the geometry of the universe , e.g. @xcite .
+ large , well defined and statistically complete samples of galaxy clusters ( which are dynamically well evolved and for which masses are approximately known ) are obvious prerequisites for such studies .
a substantial progress in the field requires samples of tens to hundreds of thousands of clusters .
surveys at several wavelengths are used or planned to be used to achieve this goal .
more in general , in x - ray surveys , galaxy clusters are detected by the radiation of the hot intracluster medium and x - ray observations are still the most efficient tools to select clean cluster samples . in addition
x - ray observations provide accurate estimates of the cluster physical parameters .
indeed the x - ray luminosity is tightly correlated to the gravitational mass , temperature and core radius @xcite .
therefore most cosmological studies involving galaxy clusters are based on x - ray surveys ( e.g. @xcite , @xcite , @xcite , @xcite .
: the psf hew as function of the offaxis angle in pointing mode .
@xmath9 : the psf hew within an encircled off - axis angle during a scan .
basically the value at 30@xmath10 , is the cumulative psf in the survey . ]
the mirror replication technique was developed for xmm - newton and has then been applied to the small satellite mission abrixas , which had scaled the xmm - newton telescopes down by a factor of about 4 .
the abrixas optical design and manufacturing process are adopted for erosita partially because the inner 27 mirror shells and therefore the focal length are kept the same .
the mirror system consists of 7 mirror modules with 54 mirror shells each and a x - ray baffle in front of each module . unlike on abrixas ,
the seven optical axes are co - aligned . compared to a large single mirror system ,
the advantages of a multiple mirror system are : shorter focal length ( reduced instrumental background ) and reduced pileup when observing bright sources .
this configuration allows a more compact telescope and multiple but identical cameras which automatically provides a 7-fold redundancy .
the capabilities of the x - ray mirror system are described by effective area , vignetting function , and psf .
the production of the flight mirrors has already started .
the erosita - ccd @xcite have 384 @xmath11 384 pixels or an image area of 28.8 mm @xmath11 28.8 mm , respectively , for a field of view of 1.03@xmath4 diameter .
the 384 channels are read out in parallel .
the nominal integration time for erosita will be 50 msec .
the integrated image can be shifted into the frame store area by less than 100 msec before it is read out within about 5 msec .
ccd together with the two camex and the ( passive ) front - end electronics are integrated on a ceramic printed circuit board ( ccd - module ) and is connected to the `` outer world '' by a flexlead . the flight - ccds have already been fabricated . for operation
the ccds have to be cooled down to -80 @xmath4c by means of passive elements ( heatpipes and radiator ) .
fluorescence x - ray radiation generated by cosmic particles is minimized by a graded shield consisting of aluminum and boron carbide . for calibration purposes ,
each camera housing contains a radioactive fe@xmath12 source and an aluminum target providing two spectral lines at 5.9 kev ( mn - k@xmath13 ) and 1.5 kev ( al - k@xmath13 ) .
the mechanism ( `` filter wheel '' ) for moving the calibration source into and out of the field of view is designed and qualified .
also the telescope structure is also qualified .
the optical bench connects the mirror system and the baffles on one side with the focal plane instrumentation on the other side .
additionally it forms the mechanical interface to the s / c bus .
the flight model manufacturing is ongoing .
the dimensions of the telescope structure is of the order 1.9 m diameter x 3.2 m height .
the total weight of erosita is 735 kg @xcite .
the instrument design is shown in figure [ grasp ] .
the 5@xmath14 point - source sensitivity vs. exposure in the 0.52 kev ( @xmath15 ) and 210 kev ( @xmath16 ) energy bands by assuming a psf hew of 30@xmath3 and 40@xmath3 , respectively .
the dashed lines is the sensitivity achieved with an average psf - hew of 25@xmath3 . @xmath6 a simulation of a 3ks observation of a [email protected]@xmath4 of erosita in survey mode .
the simulations includes cosmis+particle background , randomly distributed agn and clusters extracted from hydrodinamical simulations.,title="fig : " ] the 5@xmath14 point - source sensitivity vs. exposure in the 0.52 kev ( @xmath15 ) and 210 kev ( @xmath16 ) energy bands by assuming a psf hew of 30@xmath3 and 40@xmath3 , respectively .
the dashed lines is the sensitivity achieved with an average psf - hew of 25@xmath3 . @xmath6 a simulation of a 3ks observation of a [email protected]@xmath4 of erosita in survey mode .
the simulations includes cosmis+particle background , randomly distributed agn and clusters extracted from hydrodinamical simulations.,title="fig : " ] figure [ grasp ] shows the grasp of erosita , i.e. the product of effective area and solid angle of the field of view .
the effective area of erosita is about twice that of one xmm - newton telescope in the energy band below 2kev , whereas it is three times less at higher energies .
this is a consequence of the small f - ratio ( focal length vs. aperture ) of the erosita mirrors .
an advantage of the short focal length is also larger field of view .
the erosita angular resolution is 15 arcsec on - axis . due to the unavoidable off - axis degradation of a wolter - i telescope
, the angular resolution averaged over the field of view is of the order of 28@xmath3 ( fig .
2 ) . we will scan the entire sky for four years ( rosat 1/2 year ) .
therefore the erosita sensitivity during this all - sky survey will be approximately 30 times rosat . with the current scanning strategy
, we expect an average exposure of @xmath13 ks in the all - sky survey , with two deep fields at the ecliptic poles with an exposure of the order of 20 - 40 ks , depending on the actual mission strategy .
we have performed simulations of the radiation environment in l2 and determined , by including the cosmic components , a background intensity of 5.63 cts s@xmath18 deg@xmath19 and 3.15 cts s@xmath18 deg@xmath19 in the 0.52 kev and 210 kev energy bands , respectively .
the 0.5 - 2 kev flux limit for clusters will be , on average , of the order of 3@xmath1110@xmath20 erg cm@xmath19 s@xmath18 and 5@xmath1110@xmath22 erg cm@xmath19 s@xmath18 in the all - sky survey and in the ecliptic poles , respectively . in figure 3 we plot the erosita 5@xmath14 point source flux limit of the survey in the 0.52 kev and 210 kev energy bands as function of the exposure time .
in the all - sky survey the typical flux limit will be @xmath110@xmath20 erg cm@xmath19 s@xmath18 and @xmath13@xmath1110@xmath23 erg cm@xmath19 s@xmath18 in the 0.5 - 2 kev and 210 kev energy band , respectively . at the poles we expect to reach flux limits of the order of @xmath12@xmath1110@xmath22 erg cm@xmath19 s@xmath18 and @xmath13@xmath1110@xmath20 erg cm@xmath19 s@xmath18 in the 0.5 - 2 kev and 210 kev energy band , respectively .
note that the observation will be photon limited up to exposures of @xmath120 ks . in the 0.52 kev band
, the confusion limits of 1 source every 10 beams will be reached in about 20 - 30 ks . at this
fluxes the x - ray sky is dominated by clusters and agn , which can be separated with an angular resolution of 25@xmath330@xmath3 .
the logn - logs of clusters is well known to the proposed depth ( @xcite , @xcite , @xcite ) .
the proposed survey will identify 50,000 100,000 clusters depending on the capabilities in disentangle moderately - low extended sources from agn . concerning the number of agn we can use the logn - logs measurement in moderately wide field surveys , like xmm - cosmos @xcite , @xcite , to predict the detection 3 - 10@xmath1110@xmath24 sources , up to z@xmath17 - 8 , depending on the detection threshold .
a simulation of a 3 ks erosita observation of a typical extragalactic field is shown in figure 3 .
multi - band optical surveys to provide the required photometric and spectroscopic redshifts are already in the planning stages , and will be contemporaneous with or precede our survey .
the cluster population will essentially cover the redshift range z = 0 - 1.3 and will reveal all evolved galaxy clusters with masses above @xmath25 up to redshifts of 2 .
above this mass threshold the tight correlations between x - ray observables and mass allow direct interpretation of the data.this sample size is necessary for example to precisely characterize the cluster mass function and power spectrum in at least ten redshift bins , to follow the growth of structure with time .
meidinger , n. , andritschke , r. , ebermayer , s. , elbs , j. , hlker , o. , hartmann , r. , herrmann , s. , kimmel , n. , predehl , p. , schchner , g. , soltau , h. , strder , l. , tiedemann , l. , `` uv , x - ray , and gamma - ray space instrumentation for astronomy xvi '' .
siegmund , o.h .
proc . of the spie 7435 , pp . 743502 ( 2009 )
predehl , p. , andritschke , r. , bornemann , w. , bruninger , h. , briel , u. , brunner , h. , burkert , w. , dennerl , k. , eder , j. , freyberg , m. , friedrich , p. , f?rmetz , m. , hartmann , r. , hartner , g. , hasinger , g. , herrmann , s. , holl , p. , huber , h. , kendziorra , e. , kink , w. , meidinger , n. , mller , s. , pavlinsky , m. , pfeffermann , e. , roh ? , c. , santangelo , a. , schmitt , j. , schwope , a. , steinmetz , m. , strder , l. , sunyaev , r. , tiedemann , l. , vongehr , m. , wilms , j. , erhard , m. , gutruf , s. , jugler , d. , kampf , d. , graue , r. , citterio , o. , valsecci , g. , vernani , d. , zimmerman , m. , `` uv , x - ray , and gamma - ray space instrumentation for astronomy xv '' .
siegmund , o.h .
proc . of the spie 6686 , 668617 ( 2007 ) | erosita ( extended roentgen survey with an imaging telescope array ) is the core instrument on the russian spektrum - roentgen - gamma ( srg ) mission which is scheduled for launch in late 2012 .
erosita is fully approved and funded by the german space agency dlr and the max - planck - society .
the design driving science is the detection of 50 - 100 thousands clusters of galaxies up to redshift @xmath01.3 in order to study the large scale structure in the universe and test cosmological models , especially dark energy .
this will be accomplished by an all - sky survey lasting for four years plus a phase of pointed observations .
erosita consists of seven wolter - i telescope modules , each equipped with 54 wolter - i shells having an outer diameter of 360 mm .
this would provide and effective area at 1.5 kev of @xmath1 1500 @xmath2 and an on axis psf hew of 15@xmath3 which would provide an effective angular resolution of 25@xmath3 - 30@xmath3 . in the focus of each mirror module
, a fast frame - store pn - ccd will provide a field of view of 1@xmath4 in diameter for an active fov of @xmath10.83 deg@xmath5 . at the time of writing the instrument development
is currently in phase c / d .
address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr astrophysik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr plasma physik , d-85741 garching , germany address = institut fr astronomie und astrophysik , abteilung astronomie , universitt tbingen , sand 1 , 72076 tbingen , germany address = universit"at erlangen / nrnberg , dr .- remeis - sternwarte bamberg , sternwartstrasse 7 , d- bamberg address = astrophysikalisches institut potsdam , an der sternwarte 16 , d-14482 potsdam , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address= space research institute moscow , russian federation address = universitt hamburg , hamburger sternwarte , gojenbergsweg 112 , 21029 hamburg , germany address = institut fr astronomie und astrophysik , abteilung astronomie , universitt tbingen , sand 1 , 72076 tbingen , germany address = universitt hamburg , hamburger sternwarte , gojenbergsweg 112 , 21029 hamburg , germany address = astrophysikalisches institut potsdam , an der sternwarte 16 , d-14482 potsdam , germany address = astrophysikalisches institut potsdam , an der sternwarte 16 , d-14482 potsdam , germany address = max - planck - institut fr extraterrestrische physik , d-85741 garching , germany address = max - planck - institut fr astrophysik , d-85741 garching , germany address = institut fr astronomie und astrophysik , abteilung astronomie , universitt tbingen , sand 1 , 72076 tbingen , germany |
the ultimate goal of ultra - relativistic heavy - ion collision experiments is to test the predictions of quantum chromodynamics ( qcd ) , which is unanimously believed to be the theory of strong interactions [ 1 ] .
the ultra - relativistic nuclear collisions reveal the nature of hadronic interactions at very short time and/or distance and throw light on the role played by the internal structure of hadrons in the multiparticle production .
intensive theoretical efforts have also continued for more than three decades to understand the particle production mechanism in the nucleus - nucleus interactions .
hadronic multiplicities and their correlations can reveal information on the nature , composition , and size of the fireball from which they are originating [ 2 ] .
of particular interest is the limit up to which the chemical / thermal equilibration is possible in the fireball .
the appearance of quark gluon plasma ( qgp ) which involves a partonic medium at local thermodynamic equilibrium , and its subsequent hadronization after the phase transition , should in general drive hadrons also towards chemical equilibrium .
the theoretical description of particle production depends essentially on two ingredients : many - body interactions and the understanding of the production processes . the search for some regularities and systematics in the multiplicity distributions in hadron - hadron , hadron - nucleus and nucleus - nucleus collisions is a fascinating topic because it hints at the underlying production mechanism . however , our understanding of the subject is still very poor partially because of the reason that in soft hadronic or nuclear collisions , the role of perturbative qcd is less clear .
several new experimental information on multiparticle production have accumulated in recent years .
consequently several theoretical approaches have been developed in an attempt to understand or organise the data .
several efforts [ 3 - 8 ] , which are based on hydrodynamic descriptions , have also been proposed to explain the enormous experimental data coming from a variety of collision experiments ranging from very low energy ( e.g. , sis ) to very high energy ( e.g. , lhc ) .
in addition to this , several approaches based on statistical production of particles have recently appeared in the literature [ 9 - 23 ] .
the statistical models are based on the assumption of local filling of available phase space according to statistical laws . furthermore , there are models which try to explain the rather complicated heavy ion colliding processes by analyzing them in terms of the basic quark - gluon interaction processes e.g. , additive quark model ( aqm ) [ 24 ] , dual parton model ( dpm ) [ 25 ] or colour neutralization model ( cnm ) [ 26 ] etc .
the main difference in these models lies in the assumption about the potential number of participant partons which can interact independently .
traditional aqm treats the elastic scattering of two hadrons at high energy in terms of pomeron exchange between two quarks involving one from each hadron i.e. , the number of participants is limited to the valance quarks only [ 24 , 27 ] .
however , in dpm and cnm , the number of participant quarks is unlimited because these can involve the sea quarks as well .
many attempts have also been made in the past for searching certain systematics or scaling relations which are universal to all types of reactions , i.e. , lepton - hadron , hadron - hadron , hadron - nucleus and nucleus - nucleus collisions . any distinct deviation from these relations observed in the ultrarelativistic nuclear collisions will be a potential indicator of a new and exotic phenomenon occuring there [ 1 ] . in order to find out whether deviations arise from the presence of some exotic phenomena , e.g. , formation of qgp , where we must know the background arising from nucleus - nucleus collisions without the presence of any such phase transition [ 1 ] . in this paper
, we propose a model for the multiparticle production and search hints for deviations signaling an exotic phenomenon by comparing our results with the experimental data coming from a variety of heavy - ion collision experiments .
our model involves a main assumption that the nucleus - nucleus collision can be considered as a superposition of independent quark - quark collisions , the number of which is determind by the geometry of the collisions .
bascially in qcd , the interaction mechanism between target and projectile nuclei can be described as follows : a projectile quark exchanges a gluon with a target quark and colour forces are thus stretched between them as well as other constituents because they try to restore the colour singlet behaviour .
when two quarks thus separate , the colour force builds up a field between them and as the energy in the colour field increases , the colour tubes break up into hadrons and quark - antiquark pairs are created . in this paper , we have basically used the model given by singh et .
al . [ 28 ] and obtain the charged hadron multiplicities for @xmath6 , @xmath8 , and @xmath3 collisions .
this model considers a multiple collision scheme in which a valance quark of the incident nucleon suffers one or more inelastic collisions with a valance quark of the target nucleon .
the quark thus loses energy and momenta and produces hadrons in each quark - quark collision .
however , all the collisions are independent and their effects should be incoherently superimposed . further in @xmath3 collisions , one has to adequately incorporate the number of participating quarks and mean number of quark - quark collisions .
the rest of the paper is organized as follows : in section ii , we will give the detailed description of the model used in this paper . in section iii
, we give the results and the discussion and finally in section iv , we will give the conclusions based on this work and the future prospects of this study .
if we assume a universal mechanism of charged particle production in the hadron - hadron , hadron - nucleus and nucleus - nucleus collisions , it must be driven by the available amount of energy involved for the secondary production , and it must depend on the mean number of participant quarks .
the main ingredients of our model is based on a phenomenological model proposed earlier by singh et .
al . [ 28 ] .
let us attempt first to understand the basic particle production mechanism in @xmath6 collisions .
there are several papers emphasizing various types of fitting parametrizations in order to provide a unified description for the produced charged particles in @xmath6 collisions at various energies .
we have taken some of these functions with the values of their constants from the literature and fitted them with the experimental data .
all these parametrizations fit the experimental data for @xmath6 collisions at intermediate and higher energies but usually show some disagreement with the experimantal data at lower energies ( see fig .
1 ) . therefore
, we propose here a new parametrization to accomodate entire @xmath6 experimental data [ 29 - 38 ] from low energies to very high energies ( i.e. , from 6.15 gev upto 7 tev ) , in a unified way as follows : @xmath9 in eq .
( 1 ) , @xmath10 is the leading particle effect and @xmath11 is the available center - of - mass energy i.e. , @xmath12 , where @xmath13 is the mass of projectile and @xmath14 the mass of the target nucleon , respectively ; @xmath15 , @xmath16 , @xmath17 and @xmath18 are constants .
we find the values @xmath19 , @xmath20 , @xmath21 and @xmath22 as derived from the best fit to the data [ 29 - 38 ] .
now , we can extrapolate this expression for the produced charged particles in the hadron - nucleus interaction based on two basic assumptions .
first assumption is that the number of constituent quarks which participated in hadron - nucleus collisions share the total available center - of - mass energy @xmath23 and thus the energy available to each interacting quark becomes @xmath24 in @xmath8 collisions .
now , since @xmath25 is the available center - of - mass energy in @xmath26 collision for one effective collision of quarks of hadronic beam in the target , therefore , second assumption is that the total available square of center - of - mass energy @xmath27 in @xmath8 case becomes @xmath28 provided each quark suffers on average @xmath29 collisions .
thus , the expression for average charged hadron multiplicity in @xmath8 collisions can be expressed as follows [ 28 ] : @xmath30-\alpha.\ ] ] in the above eq .
( 2 ) , @xmath31 , where @xmath29 is the mean number of inelastic quark collisions in the target nucleus and is defined as : @xmath32 . here
@xmath33 is the mass number of the target nucleus and @xmath34 is the quark - nucleon inelastic interaction cross - section and equal to one - third of nucleon - nucleon inelastic cross - section ( @xmath35 ) i.e. , @xmath36 [ 17 ] .
also @xmath37 , the quark - nucleus inelastic interaction cross - section and is obtained from glauber s approximation by using the following expression [ 18 ] : @xmath38,\ ] ] where profile function @xmath39 is related to nuclear density , @xmath40 by the relation : @xmath41 we use the following expression to calculate the nuclear density : @xmath42 where @xmath43 and @xmath44 are constants for any nucleus having mass number @xmath33 , @xmath45 is the normalization constant and @xmath46 is the impact parameter .
we have taken the values of the constants from the ref . [
furthermore , @xmath47 in eq .
( 3 ) , is the mean number of participant quarks and is defined as follows : @xmath48 where @xmath49 is the number of valance quarks in the nucleus a. + the generalization of the above picture for the case of nucleus - nucleus collisions goes along the same line and can be given as follows : @xmath50,\ ] ] where @xmath51 and the mean number of inelastic quark collision @xmath52 can be given as follows : @xmath53 furthermore , mean number of participating quarks @xmath54 can be calculated by generalizing eq .
( 6 ) in the following manner : @xmath55,\ ] ] where @xmath56 is the inelastic cross - section for nucleus a- nucleus b collision and can be expressed in the following manner [ 18 ] : @xmath57 ^ 2,\ ] ] where c is a constant and has a value 4.45 for nucleus - nucleus collisions .
one can see that the parametrization as given by eq .
( 7 ) gives the most general relation relating nucleus - nucleus collisions to hadron - nucleus and hadron - proton collisions and the values of the parameters @xmath58 , and @xmath18 remain unaltered which shows the similarity in the role of basic quark - gluon interaction in all these processess . in a search for creating quark gluon plasma ( qgp ) , greater emphasis
is laid on the central or head - on collisions of two nuclei . in such a case
, we assume that all the quarks of the beam nucleus are wounded and the resulting mean multiplicity can be obtained by using eq .
( 7 ) in the following manner : @xmath59.\ ] ] the pseudorapidity distribution of charged particles is another very important quantity in the studies of particle production mechanism in high energy @xmath60 and @xmath3 collisions .
it has been pointed out that @xmath61 can be used to obtain the temperature ( @xmath62 ) and density ( @xmath63 ) of the qgp [ 40 - 42 ] . moreover , mid - rapidity density reflects the different stages of the reaction . to calculate the pseudorapidity density of charged hadrons
, we first fit the experimental data of @xmath5 by using similar parametrization as was used in eq .
( 1 ) but with slightly different values of constants
. further extrapolation of two component model [ 43 - 45 ] gives @xmath64 as follows : @xmath65,\ ] ] where @xmath66 quantifies the relative contribution from hard and soft processes .
the fraction @xmath66 corresponds to the hard processes and the remaining fraction ( @xmath67 ) arising from soft processes .
thus the contribution from hard processes is propotional to the total number of collisions i.e. , @xmath68 and the portion from soft processes is propotional to the participating quark constituents i.e. , @xmath69
. collision at various @xmath7 . ]
collisions at various @xmath7 . ] in fig . ( 1 ) , we present the inelastic ( filled symbols ) and non - single diffractive ( nsd ) data ( open symbols ) of charged hadron multiplicity in full phase space for @xmath6 collisions at various center - of - mass energies from different experiments e.g. , isr , ua5 and e735 [ 29 - 38 ] . we use inelastic data at very low energies ( filled symbols ) because there is no nsd data available for these energies and also the trend shows that the difference between inelastc and nsd data will be very small at lower energies .
further , we fit this data set with four different functional forms .
the short - dashed line has the functional form as : @xmath70 which is actually inspired by the fermi - landau model [ 46 , 47 ] .
it provides a resonable fit to the data at higher @xmath7 with @xmath71 and @xmath72 [ 29 ] .
however , since @xmath44 summarizes the leading particle effect , it should not be much larger than two .
the dotted line has the functional form as : @xmath73 @xmath74 @xmath75 and it fits the data well at higher @xmath7 but shows a disagreement with the experimental data at lower center - of - mass energies .
the dashed - dotted line represents the form @xmath76 and it also provides a good qualitative description of the data with @xmath77 and @xmath78 [ 29 ] .
the solid line represents our parametrization given by eq .
( 1 ) and it undoubtedly represents the most reasonable good fit to the data starting from very low upto very high @xmath7 . . ]
2 , shows the variation of mean multiplicity of charged hadrons produced in central @xmath79 collision with respect to @xmath7 .
we also compare our model results ( solid line ) and the modified glauber model results ( dash - dotted line ) [ 48 ] with the experimental data of ags and rhic [ 49 - 52 ] .
we would like to mention here that the main difference in modified glauber model with respect to the standard glauber calculations is that for each nucleon - nucleon collision modified model uses the value of @xmath80 at the corresponding centre - of - mass nucleon - nucleon collision energy .
we find that the results obtained from our model give a good description to the experimental data in comparison to the modified glauber model predictions , especially at lower center - of - mass energies . in fig .
3 , we show the variation of mean - multiplicity calculated from our model for @xmath79 , @xmath81 and @xmath82 central collisions with respect to @xmath7 .
we also compare our model results with the experimantal data of rhic and sps experiments [ 49 - 53 ] .
one can see from fig .
3 , that the model results give excellent fit to the separate experimental data points for @xmath82 , @xmath79 and @xmath81 , respectively .
this concludes that our phenomenological model works well because it considers the picture of incoherent superposition of basic quark - quark collisions in describing the nucleus - nucleus collisions . in table
i , we have shown the mean multiplicity of charged hadrons in minimum - bias as well as in central @xmath83 collision at @xmath84 gev .
we have also shown the experimental results [ 54 ] for this collision for comparison .
the good agreement between the experimental and our results shows that our model can be used for asymmetric colliding nuclei as well .
collisionat various @xmath7 .
filled symbols are data from inelastic @xmath6 events [ 32 , 53 - 54 ] and open symbols are experimental data for nsd events [ 33 - 36 , 58 - 59 ] .
solid line is the outcome of our parametrization . ] in fig .
4 , we present the inelastic ( filled symbols ) and non - single diffractive ( nsd ) data ( open symbols ) of @xmath1 for @xmath6 collisions at various center of mass energies from different experiments e.g. , isr , ua5 , e735 , rhic and lhc [ 32 , 33 - 36 , 55 - 59 ] . further , we fit ( solid line ) this data set with our functional form of eq .
( 1 ) and obtained the values of constants as @xmath85 , @xmath86 , @xmath87 and @xmath88 .
the central rapidity hadron density thus continues to increase with the colliding energy at lhc in a triple logarithmic form .
collisionat various @xmath7 .
symbols are data from various heavy ion collision experiments [ 49 - 53 , 60 - 61 ] . ] in fig . 5
, we plot the variation of @xmath1 obtained in our model for @xmath79 , @xmath81 and @xmath82 central collisions with respect to @xmath7 .
we also compare our model results with the experimental data of rhic , sps and lhc experiments [ 49 - 53 , 60 - 61 ] .
our model results compare well with the experimental data from very low energy experiments to very high energy experiments . as we did nt take any effect of final - state interactions in our model , it verifies that the hadron multiplicity in @xmath3 collisions is mainly driven by initial parton production and the effect of final - state interaction is negligibly small as suggested in some earlier models [ 62 - 63 ] . in table.ii , we give the predictions for the mean multiplicity of charged hadrons , produced in minimum - bias events as well as in central events , based on our model calculations at lhc and cbm energies . specifically , we calculate the charged hadron multiplicity in @xmath81 collisions at @xmath89 tev for lhc and the mean multiplicity of charged hadrons in @xmath79 collisions at @xmath90 agev and @xmath91 agev for future cbm experiment .
we also compare our model results with the other model predictions like hadron string dynamics ( hsd ) model predictions for cbm experiment [ 64 - 65 ] and model prediction of w. busza for lhc experiment [ 66 - 67 ] .
we have given a parametrization which correctly describes the multiplicity distributions in particle and nuclear collisions from a few gev upto the highest lhc energies .
it involves the interactions of the constituent quarks of the colliding beams or beam and target .
the collision of the two lorentz - contracted particles results in a complete thermalized system .
we assume that the production of secondary particles is proportional to the fraction of the available energy for the participant quarks .
this picture describes consistently the ` soft ' hadron production in @xmath6 , @xmath92 , @xmath3 and @xmath93 collisions . in heavy - ion collisions , more than one quark per nucleon interacts due to the large size of the nucleus as well as due to a large travel path to be travelled inside the nucleus [ 68 ] . for more central collisions ,
larger number of interactions occur and hence we get larger energy available for the secondary particle production . in the most central nucleus - nucleus collisions , all three constituent quarks from each nucleon will interact simultaneously and they deposit their energy into the thermalized collisions volume .
we find that our parametrization gives the excellent fit to the @xmath6 data for the entire energy range .
we also find that a suitable extension of this description works very well for the @xmath3 collisions again for the entire energy range .
we have also compared our model results with the results obtained from modified glauber model to demonstrate the difference between nucleon - nucleon interaction and quark - quark interaction pictures .
we have shown the charged particle multiplicity in central as well as in minimum - bias @xmath83 collisions at @xmath84 gev and we find it compares well with the experimental result .
this exercise shows that our model works well for assymetric colliding nuclei case also .
we have also proposed an extension of the two - component model based on our model and attempted to get the rapidity distribution and we find that it again works well .
we have given our prediction for lhc and cbm data as well and compare them with some recent model predictions [ 64 - 67 ] .
in conclusion , we have attempted to draw certain universal mechanism for ` soft ' hadron production in hadron - hadron , hadron - nucleus and nucleus - nucleus collisions .
we emphasized that qcd does not work for these cases .
so lacking a workable theory , we have relied on the phenomenology and all the salient features of our model fit very well with the experimental data .
we hope that the work done here will throw enough light on the multiparticle production mechanism in nucleus - nucleus collisions .
this also reveals that any deviations observed in the data from the predictions will help us in identifying qgp formation .
furthemore , we expect that this model can also be extended to accomodate the rapidity , pseudorapidity as well as transverse mass distribution of charged hadrons in the whole rapidity region which will be studied in a future publication .
+ * acknowledgments * + ak and pks are grateful to the university grants commission ( ugc ) , new delhi for providing a research grant .
bks also sincerely acknowledges the indian space research organization ( isro ) , india for providing the financial support .
99 c. p. singh , phys .
236 , 147 ( 1993 ) .
p. braun - munzinger , k. redlich and j. stachel , quark gluon plasma 3 , eds .
r. hwa , x .-
wang ( world scientific , singapore 2004 ) .
j. d. bjorken , phys . rev .
d 27 , 140 ( 1983 ) .
p. f. kolb , u. w. heinz , p. huovinen , k. j. eskola , and k. tuominen , nucl .
phys . a 696 , 197 ( 2001 ) .
d. teaney , j. lauret , and e. v. shuryak , arxiv : nucl - th/0110037 .
p. f. kolb , u. w. heinz , arxiv : nucl - th/0305084 .
c. nonaka , e. honda , and s. muroya , eur .
j. c 17 , 663 ( 2000 ) .
t. hirano , k. morita , s. muroya , and c. nonaka , phys .
c 65 , 061902 ( 2002 ) .
j. cleymans , and k. redlich , phys . rev .
81 , 5284 ( 1998 ) .
p. braun - munzinger , d. magestro , k. redlich and j. stachel , phys .
b 518 , 41 ( 2001 ) and reference therein . p. braun - munzinger , i. heppe , and j. stachel , phys .
b 465 , 15 ( 1999 ) .
f. becattini , j. cleymans , a. keranen , e. suhonen , and k. redlich , phys . rev .
c 64 , 024901 ( 2001 ) .
m. gazdzicki et .
al . , z. phys .
c 65 , 215 ( 1995 ) ; acta .
b 30 , 2705 ( 1999 ) . a. tounsi and k. redlich , eur .
j. c 24 , 529 ( 2002 ) ; j. phys .
g 28 , 2095 ( 2002 ) . j. sollfrank et .
al . , phys .
c 55 , 392 ( 1997 ) .
p. huovinen et .
al . , phys .
b 503 , 58 ( 2001 ) . n. xu , m. kaneta , nucl .
a 698 , 306 ( 2002 ) ; m. kaneta and n. xu , j. phys .
g 27 , 589 ( 2001 ) .
w. broniowski and w. florkowski , phys .
lett 87 , 272302 ( 2001 ) .
k. redlich , nucl .
a 698 , 94 ( 2002 ) .
g. d. yen , m. i. gorenstein , w. greiner , s. n. yang , phys .
c 56 , 2210 ( 1997 ) t. s. biro , p. levai , j. zimanyi , j. phys .
g 28 , 1561 ( 2002 ) .
braun - munzinger , and j. stachel , nucl .
a 690 , 119c ( 2001 ) .
a. andronic , p. braun - munzinger , and j. stachel , arxiv : nucl - th/0511071v3 .
v. v. anisovich and v. m. shektar , nucl .
b55 , 455 ( 1973 ) ; a. bialas et .
al . , acta phys .
b8 , 585 ( 1977 ) .
a. capella and j. tranthanhvan , phys .
b93 , 146 ( 1980 ) ; a. capella , j. tranthanhvan , and j. kwiecinski , phys .
58 , 2015 ( 1987 ) . s. j. brodsky , j. f. gunion , j. h. kuhn , phys . rev .
39 , 1120 ( 1977 ) .
p. desgrolard , m. giffon , e. martynov , e. predazzi , eur .
j. c9 , 623 ( 1999 ) .
c. p. singh , m. shyam , s. k. tuli , phys .
c40 , 1716 ( 1989 ) ; m. shyam , c. p. singh and s. k , tuli , phys .
b164 , 189 ( 1985 ) ; c. p. singh , and m. shyam , phys .
b171 , 125 ( 1986 ) .
jan fiete grosse - oetringhaus , and klaus reygers , j. phys .
g 37 , 083001 ( 2010 ) .
a. breakstone et al .
d30 , 528 ( 1984 ) .
p. slattery , phys .
29 , 1624 ( 1972 ) .
w. thome et .
al . , nucl .
b129 , 365 ( 1977 ) .
g. arnison et .
b123 , 108 ( 1983 ) ; c. albajar et .
b 335 , 261 ( 1990 ) . r. e. ansorge et .
al . , z. phys .
c43 , 357 ( 1989 ) . g. j. alner et .
al . , phys .
b160 , 193 ( 1985 ) ; phys .
154 , 247 ( 1987 ) .
f. abe et .
al . , phys . rev .
d41 , 2330 ( 1990 ) .
j. whitmore , phys .
10 , 273 ( 1974 ) .
wang , ph.d .
thesis , iowa state university ( 1991 ) .
h. devries , c. w. dejager , and c. devries , atomic data and nuclear data tables 36 , 495 ( 1987 ) .
x. n. wang , and m. gyulassy , phys .
86 , 3496 ( 2001 ) .
a. capella and d. sousa , phys .
b 511 , 185 ( 2001 ) .
k. j. eskola , k. kajantie , p. v. ruuskanen , and k. tuominen , nucl .
b 570 , 379 ( 2000 ) .
g. n. fowler , e. m. friedlander , r. m. weiner , and g. wilk , phys .
rev . lett .
57 , 2119 ( 1986 ) .
l. grandchamp and r. rapp , phys .
b523 , 60 ( 2001 ) ; nucl .
phys . a 709 , 415 ( 2002 ) .
x. n. wang and m. gyulassy , phys .
d 44 , 3501 ( 1991 ) ; d. kharzeev , and m. nardi phys .
b 507 , 121 ( 2001 ) ; w .- t .
deng , x .-
n . wang , and r. xu , phys .
c 83 , 014915 ( 2011 ) .
e. fermi , prog .
phys . 5 , 570 ( 1950 ) .
wong , phys .
c78 , 054902 ( 2008 ) . g. feofilov and a. ivanov , journal of physics : conference series 5 , 230 ( 2005 ) .
b. alver et .
al . , phys .
c 83 , 024913 ( 2011 ) .
s. s. adler et .
al . , phys .
c 71 , 034908 ( 2005 ) . b. b. back et .
al . , phys .
c 74 , 021902 ( r ) ( 2006 ) .
j. l. klay et .
c 68 , 054905 ( 2003 ) .
s. v. afanasiev et .
al . , phys .
c 66 , 054902 ( 2002 ) . b. b. back et .
( phobos coll .
c72 , 031901 ( 2005 ) .
sun , f .- h .
liu , e .- q .
wang , y. sun , and z. sun , phys .
c 83 , 014001 ( 2011 ) .
r. nouicer et .
al . , j. phys .
g 30 , s1133 ( 2004 ) . b. i. abelev et .
al . , phys .
c 79 , 034909 ( 2009 ) .
k. aamodt et .
al . , eur .
j. c 68 , 89 ( 2010 ) ; eur .
j. c 65 , 111 ( 2010 ) .
v. khachatryan et .
al . , j. high energy phys . 2010 , 02041 ( 2010 ) ; phys .
105 , 022002 ( 2010 ) .
k. aamodt et .
al . , phys .
106 , 032301 ( 2011 ) .
s. chatrchyan et .
al . , j. high energy phys .
08 , 141 ( 2011 ) .
k. j. eskola , v. j. kolhinen , and c. a. salgado , eur .
j. c 9 , 61 ( 1999 ) .
z. w. lin , s. pal , c. m. ko , b. a. li , and b. zhang , phys . rev .
c 64 , 011902 ( 2001 ) .
cbm report 2012 - 01 , nuclear matter physics at sis-100 , http://www.gsi.de/documents/doc-2011-aug-29-1.pdf w. cassing , e. l. bratkovskaya , s. juchem , nucl .
phys . a 674 , 249 ( 2000 ) .
w. busza , acta .
b 35 , 2873 ( 2004 ) .
n. armesto et .
g 35 , 5400 ( 2008 ) .
edward k. g. sarkisyan , alexander s. sakharov , eur .
j. c 70 , 533 ( 2010 ) . | a systematic study of charged hadron multiplicities ( @xmath0 ) at various collision energies is very much important in understanding the basic production mechanism of the hadrons in nucleus - nucleus collision experiments .
furthermore , the variations of @xmath0 in nucleus - nucleus collisions with respect to the colliding energy and mass number can provide a potential probe for the formation of quark gluon plasma ( qgp ) in the laboratory . in this paper , we propose a phenomenological model based on the constituent quark - quark interactions to calculate the average multiplicity ( @xmath0 ) and pseudorapidity density at mid - rapidity ( @xmath1 ) of charged hadrons at various center - of - mass energies @xmath2 for nucleus - nucleus ( @xmath3 ) collisions .
we first propose a new parametrization for @xmath4 and @xmath5 in @xmath6 interactions based on some initial inputs which fit the experimental data very well .
we further extend this parametrization by using simple phenomenological assumptions regarding mean number of participating quarks and mean number of collisions to obtain the @xmath0 and @xmath1 for @xmath3 collisions and show their dependencies on the mass number of colliding nuclei as well as on @xmath7 .
we also compare the results obtained from our model with the results obtained from the modified glauber model in order to demonstrate the difference between the two formalisms . finally , we compare the charged hadron multiplicity and pseudorapidity density at mid - rapidity for @xmath3 collisions obtained from our model with the available experimetal data from various heavy - ion collision experiments and give our predictions for @xmath3 collisions at the large hadron collider ( lhc ) and at compressed baryonic matter ( cbm ) experiments . +
pacs numbers : 12.38.mh , 24.85.+p , 25.75.-q , 25.75.dw |
we consider the ill - posed inverse problems of the form @xmath0 where @xmath1 is a bounded linear operator between two hilbert spaces @xmath2 and @xmath3 whose inner products and the induced norms are denoted as @xmath4 and @xmath5 respectively which should be clear from the context .
here the ill - posedness of ( [ 1.1 ] ) refers to the fact that the solution of ( [ 1.1 ] ) does not depend continuously on the data which is a characteristic property of inverse problems . in practical applications ,
one never has exact data , instead only noisy data are available due to errors in the measurements . even if the deviation is very small , algorithms developed for well - posed problems may fail , since noise could be amplified by an arbitrarily large factor .
therefore , regularization methods should be used in order to obtain a stable numerical solution .
one can refer to @xcite for many useful regularization methods for solving ( [ 1.1 ] ) ; these methods , however , have the tendency to over - smooth solutions and hence are not quite successful to capture special features . in case a priori information on the feature of the solution of ( [ 1.1 ] ) is available , we may introduce a proper , lower semi - continuous , convex function @xmath6 $ ] such that the sought solution of ( [ 1.1 ] ) is in @xmath7 . by taking @xmath8 and @xmath9 , the solution of ( [ 1.1 ] ) with the desired feature
can be determined by solving the constrained minimization problem @xmath10 where @xmath11 denotes the bregman distance induced by @xmath12 at @xmath13 in the direction @xmath14 , i.e. @xmath15 when only a noisy data @xmath16 is available , an approximate solution can be constructed by the tikhonov - type method @xmath17 when the regularization parameter @xmath18 is given , many efficient solvers were developed to compute @xmath19 when @xmath12 is the @xmath20 or the total variation function .
unfortunately almost all these methods do not address the choice of @xmath18 which , however , is important for practical applications . in order to use these solvers
, one has to perform the trial - and - error procedure to find a reasonable @xmath18 which is time consuming . on the other hand , some iterative methods , equipping with proper termination criteria , were developed to find approximate solutions of ( [ 10.14.1 ] ) , see @xcite and references therein .
these iterative methods have the advantage of avoiding the difficulty for choosing the regularization parameter .
however in each iteration step one has to solve a minimization problem similar to ( [ tik ] ) , and overall it may take long time . in this paper
we will propose a fast iterative regularization methods for solving ( [ 10.14.1 ] ) by splitting @xmath21 and @xmath12 into different steps .
our idea is to exploit the hilbert space structure of the underlying problem to build each iterate by first applying one step of a well - established classical regularization method and then penalizing the resultant by the convex function @xmath12 . to motivate the method
, we consider the exact data case .
we take an invertible bounded linear operator @xmath22 which can be viewed as a preconditioner and rewrite ( [ 10.14.1 ] ) into the equivalent form @xmath23 the corresponding lagrangian is @xmath24 where @xmath25 represents the dual variable
. then a desired solution of ( [ 10.14.1 ] ) can be found by determining a saddle point of @xmath26 if exists .
let @xmath27 be a current guess of a saddle point of @xmath26 , we may update it to get a new guess @xmath28 as follows : we first update @xmath29 by solving the proximal maximization problem @xmath30 with a suitable step length @xmath31 .
we then update @xmath32 by solving the minimization problem @xmath33 by straightforward calculation and simplification it follows @xmath34 which is the one step result of the uzawa algorithm @xcite or the dual subgradient method @xcite , where @xmath35 and @xmath36 denote the adjoint operators of @xmath21 and @xmath37 respectively . by setting @xmath38 and @xmath39
, the above equation can be transformed into the form @xmath40 now we may apply the updating scheme ( [ 10.13.2 ] ) to @xmath41 iteratively but with dynamic preconditioning operator @xmath42 and variable step size @xmath43 .
this gives rise to the following iterative methods @xmath44 the performance of the method ( [ method0 ] ) depends on the choices of @xmath45 .
if we take @xmath46 for all @xmath47 , ( [ method0 ] ) becomes the method that has been studied in @xcite which is the generalization of the classical landweber iteration and is known to be a slowly convergent method .
in this paper we will consider ( [ method ] ) with @xmath48 for all @xmath47 , where @xmath49 is a decreasing sequence of positive numbers .
this yields the nonstationary iterative method @xmath50 observing that when @xmath51 and @xmath52 for all @xmath47 , ( [ method ] ) reduces to the nonstationary iterated tikhonov regularization @xmath53 whose convergence has been studied in @xcite and it has been shown to be a fast convergent method when @xmath49 is a geometric decreasing sequence .
this strongly suggests that our method ( [ method ] ) may also exhibit fast convergence property if @xmath49 and @xmath54 are chosen properly .
we will confirm this in the present paper .
it is worthy to point out that each iteration in ( [ method ] ) consists of two steps : the first step involves only the operator @xmath21 and the second step involves only the convex function @xmath12 .
this splitting character can make the computation much easier .
this paper is organized as follows . in section 2 , we start with some preliminary facts from convex analysis , and then give the convergence analysis of the method ( [ method ] ) when the data is given exactly . in case
the data is corrupted by noise , we propose a stopping rule to terminate the iteration and establish the regularization property .
we also give a possible extension of our method to solve nonlinear inverse problems in hilbert spaces . in section 3
we test the performance of our method by reporting various numerical simulations , including the image deblurring , the determination of source term in poisson equation and the de - autoconvolution problem .
in this section we first give the convergence analysis of ( [ method ] ) with suitable chosen @xmath55 when @xmath6 $ ] is a proper , lower semi - continuous function that is strongly convex in the sense that there is a constant @xmath56 such that @xmath57 for all @xmath58 and @xmath59 .
we then consider the method when the data contains noise and propose a stopping rule to render it into a regularization method .
our analysis is based on some important results from convex analysis which will be recalled in the following subsection . given a convex function @xmath6 $ ]
, we will use @xmath60 to denote its effective domain .
it is called proper if @xmath61 .
given @xmath62 , the set @xmath63 is called the subdifferential of @xmath12 at @xmath64 and each element @xmath65 is called a subgradient . our convergence analysis of ( [ method ] )
will not be carried out directly under the norm of @xmath2 .
instead we will use the bregman distance ( @xcite ) induced by @xmath12 .
given @xmath65 , the quantity @xmath66 is called the bregman distance induced by @xmath12 at @xmath64 in the direction @xmath67 .
it is clear that @xmath68 .
however , bregman distance is not a metric distance since it does not satisfy the symmetry and the triangle inequality in general . nevertheless , when @xmath12 is strongly convex in the sense of ( [ sc ] ) , there holds ( @xcite ) @xmath69 which means that the bregman distance can be used to detect information under the norm of @xmath2 .
although @xmath12 could be non - smooth , its fenchel conjugate can have enough regularity provided @xmath12 has enough convexity .
the fenchel conjugate of @xmath12 is defined by @xmath70 for a proper , lower semi - continuous , convex function @xmath12 , there always holds @xmath71 consequently , the bregman distance can be equivalently written as @xmath72 if in addition @xmath12 is strongly convex in the sense of ( [ sc ] ) , then @xmath73 , @xmath74 is frchet differentiable , and its gradient @xmath75 satisfies @xmath76 i.e. @xmath75 is lipschitz continuous .
these facts are crucial in the forthcoming convergence analysis and their proofs can be found in many standard textbooks , cf .
@xcite .
we consider the convergence of the method ( [ method ] ) under the condition that @xmath12 is proper , lower semi - continuous , and strongly convex in the sense of ( [ sc ] ) .
we will always assume that ( [ 1.1 ] ) has a solution in @xmath77 . by taking @xmath78 and define @xmath79 as an initial guess , we define @xmath80 to be the solution of ( [ 1.1 ] ) in @xmath77 satisfying @xmath81 it is easy to show that such @xmath80 is uniquely defined .
our aim is to show that the sequence @xmath82 produced by ( [ method ] ) eventually converges to @xmath80 if @xmath55 is chosen properly . to this end
, we first consider the monotonicity of the bregman distance @xmath83 with respect to @xmath47 for any solution @xmath84 of ( [ 1.1 ] ) in @xmath77 . by the subdifferential calculus and the definition of @xmath85 , it is easy to see that @xmath86 and hence @xmath87
. therefore , in view of ( [ 10.13.3 ] ) and ( [ 10.13.4 ] ) we have @xmath88 using the definition of @xmath89 in ( [ method ] ) and @xmath90 we obtain @xmath91 if @xmath92 , we may choose @xmath55 such that @xmath93 with @xmath94 , then it yields @xmath95 when @xmath96 , the inequality ( [ eq:11.22 ] ) obviously holds for any @xmath97 .
we observe that the @xmath55 chosen by ( [ 10.13.5 ] ) could be very large when @xmath98 is small .
using such a choice of @xmath55 it could make the method numerically unstable , in particular when the data contains noise . to avoid this
, we take a preassigned number @xmath99 and then set @xmath100 the above argument then shows the following monotonicity result .
[ lem10.13.1 ] if @xmath55 is chosen by ( [ 10.13.7 ] ) with @xmath94 and @xmath99 , then @xmath101 and @xmath102 for any solution @xmath84 of ( [ 1.1 ] ) in @xmath77 , where @xmath103 .
we will use lemma [ lem10.13.1 ] to derive the convergence of the method ( [ method ] ) . for the step size @xmath55 defined by ( [ 10.13.7 ] ) , it is easy to see that @xmath104 where we used the inequality @xmath105 to derive the left inequality .
this together with ( [ 10.3.4 ] ) implies @xmath106 since @xmath107 , we can further derive that @xmath108 the following main result shows that the method ( [ method ] ) is indeed convergent if @xmath55 is chosen by ( [ 10.13.7 ] ) .
[ thm1 ] let @xmath6 $ ] be a a proper , lower semi - continuous function that is strongly convex in the sense of ( [ sc ] ) .
if @xmath49 is a decreasing sequence of positive numbers and if @xmath55 is chosen by ( [ 10.13.7 ] ) with @xmath94 and @xmath99 , then for the method ( [ method ] ) there hold @xmath109 the proof is based on the following useful result .
[ general ] consider the equation ( [ 1.1 ] ) .
let @xmath110 $ ] be a proper , lower semi - continuous and strong convex function .
let @xmath111 and @xmath112 be such that 1 .
@xmath113 for all @xmath47 ; 2 . for any solution @xmath84 of (
[ 1.1 ] ) in @xmath77 the sequence @xmath114 is monotonically decreasing ; 3 .
4 . there is a subsequence @xmath116 with @xmath117 such that for any solution @xmath84 of ( [ 1.1 ] ) in @xmath77 there holds @xmath118 then there exists a solution @xmath119 of ( [ 1.1 ] ) in @xmath77 such that @xmath120 if , in addition , @xmath121 for all @xmath47 , then @xmath122
. * proof .
* this is a slight modification of ( * ? ? ?
* proposition 3.6 ) , we include here the proof for completeness .
we first show the convergence of @xmath123 .
for any @xmath124 we have from the definition of bregman distance that @xmath125 by the monotonicity of @xmath114 and ( [ 10.3.2 ] ) we obtain that @xmath126 as @xmath127 . in view of the strong convexity of @xmath12
, it follows that @xmath123 is a cauchy sequence in @xmath2 .
thus @xmath128 for some @xmath129 .
since @xmath130 , we have @xmath131 . in order to show @xmath132
, we use @xmath133 to obtain @xmath134 in view of ( [ 10.3.2 ] ) and @xmath128 as @xmath135 , there is a constant @xmath136 such that @xmath137 therefore @xmath138 for all @xmath139 . by using the lower semi - continuity of @xmath12 we obtain @xmath140 this implies that @xmath132 .
next we derive the convergence in bregman distance .
since @xmath141 is monotonically decreasing , the limit @xmath142 exists . by taking @xmath135 in ( [ 10.3.1 ] ) with @xmath143 and using the lower semi - continuous of @xmath12
, we obtain @xmath144 which is true for all @xmath145 .
letting @xmath146 and using ( [ 10.3.2 ] ) gives @xmath147 .
thus @xmath148 , i.e. @xmath149 . finally we show that @xmath122 .
we use ( [ 5.5.3.1 ] ) with @xmath150 replaced by @xmath80 to obtain @xmath151 because of ( [ 10.3.2 ] ) , for any @xmath152 we can find @xmath153 such that @xmath154 we next consider @xmath155 .
since @xmath121 , we can find @xmath156 such that @xmath157 .
consequently @xmath158 since @xmath159 as @xmath160 , we can find @xmath161 such that @xmath162 therefore @xmath163 for all @xmath164 .
since @xmath152 is arbitrary , we obtain @xmath165 . by taking @xmath135 in ( [ 5.5.7 ] ) and using the lower semi - continuity of @xmath12 we obtain @xmath166 . according to the definition of @xmath80
we must have @xmath167 . by uniqueness it follows @xmath122 .
+ * proof of theorem [ thm1 ] .
* we will use proposition [ general ] to complete the proof . by the definition of @xmath168 we always have @xmath169 .
it remains to verify the four conditions in proposition [ general ] . by the definition of @xmath85 we have @xmath86 which implies ( i ) in proposition [ general ] .
moreover , lemma [ lem10.13.1 ] and ( [ 10.13.9 ] ) confirm ( ii ) and ( iii ) in proposition [ general ] respectively .
it remains only to verify ( iv ) in proposition [ general ] . to this end , we consider @xmath170 in view of ( [ 10.13.8 ] ) , we have @xmath171 . moreover , by the definition of the method ( [ method ] ) , if @xmath172 for some @xmath47 , then @xmath173 for all @xmath174 .
consequently , we may choose a strictly increasing subsequence @xmath116 of integers such that @xmath175 and @xmath176 , for each @xmath177 , is the first integer satisfying @xmath178 for this chosen @xmath116 it is easy to see that @xmath179 inddeed , for @xmath180 , we can find @xmath181 such that @xmath182 and thus , by the definition of @xmath183 , we have @xmath184 . with the above chosen @xmath116 , we will show that ( [ 10.3.2 ] ) holds for any solution @xmath84 of ( [ 1.1 ] ) in @xmath77 . by the definition of @xmath185 we have for @xmath124 that @xmath186 therefore @xmath187 by using the monotonicity of @xmath49 and ( [ 9.29.1 ] )
, we have for @xmath188 that @xmath189 consequently , it follows from ( [ 10.3.4 ] ) that @xmath190 which , together with the monotonicity of @xmath114 , implies ( [ 10.3.2 ] ) .
the proof is therefore complete .
we next consider the situation that the data contains noise .
thus , instead of @xmath192 , we only have noisy data @xmath16 satisfying @xmath193 with a small known noise level @xmath194 .
the corresponding method takes the form @xmath195 with suitably chosen step length @xmath196 , where @xmath197 and @xmath198 . in order to terminate the method , we need some stopping criterion .
it seems that a natural one is the discrepancy principle @xmath199 for some number @xmath200 .
unfortunately , we can not prove the regularization property for the method terminated by the discrepancy principle ; furthermore , numerical simulations indicate that the discrepancy principle might not always produce satisfactory reconstruction result . therefore , the discrepancy principle might not be a natural rule to terminate ( [ method_noise ] ) . recall that
when we motivate our method , we consider the preconditioned equation @xmath201 instead of @xmath41 .
this indicates that it might be natural to stop the iteration as long as @xmath202 is satisfied for the first time . the stopping rule ( [ pmd ] )
can be viewed as the discrepancy principle applied to the preconditioned equation . since
the right hand side of ( [ pmd ] ) involves @xmath192 which is not available , it can not be used in practical applications .
considering @xmath203 , we may replace the right hand side of ( [ pmd ] ) by @xmath204 which leads to the following stopping rule .
[ rule1 ] let @xmath200 be a given number .
we define @xmath205 to be the first integer such that @xmath206 in the context of tikhonov regularization for linear ill - posed inverse problems , a similar rule was proposed in @xcite to choose the regularization parameter .
the rule was then generalized and analyzed in @xcite for nonlinear tikhonov regularization and was further extended in @xcite as a stopping rule for the iteratively regularized gauss - newton method for solving nonlinear inverse problems in hilbert spaces .
combining rule [ rule1 ] with ( [ method_noise ] ) and using suitable choice of the step length @xmath207 it yields the following algorithm .
[ nonstationary iterative method with convex penalty ] [ alg1 ] 1 .
take @xmath200 , @xmath208 , @xmath99 and a decreasing sequence @xmath49 of positive numbers ; 2 .
take @xmath78 and define @xmath209 as an initial guess ; 3 .
for each @xmath210 define @xmath211 and @xmath212 by ( [ method_noise ] ) , where @xmath213 4 .
let @xmath205 be the integer determined by rule [ rule1 ] and use @xmath214 as an approximate solution .
the following lemma shows that algorithm [ alg1 ] is well defined and certain monotonicity result holds along the iteration if @xmath208 is suitably small .
[ lem1 ] let @xmath6 $ ] be a a proper , lower semi - continuous function that is strongly convex in the sense of ( [ sc ] ) .
if @xmath49 is a decreasing sequence of positive numbers and @xmath207 is chosen by ( [ 10.3.6 ] ) with @xmath215 and @xmath99 , then rule [ rule1 ] defines a finite integer @xmath205 . moreover , if @xmath216 , then for the sequences @xmath217 and @xmath218 defined by ( [ method_noise ] ) there holds @xmath219 for any solution @xmath84 of ( [ 1.1 ] ) in @xmath77 .
* let @xmath220 . by using the similar argument in the proof of lemma [ lem10.13.1 ]
we can obtain @xmath221 in view of @xmath222 and the choice of @xmath207 , it follows that @xmath223 by the definition of @xmath205 and @xmath224 we have @xmath225 therefore , we have with @xmath226 that @xmath227 this shows the monotonicity result ( [ monotone ] ) and @xmath228 for all @xmath220 .
we may sum the above inequality over @xmath47 from @xmath229 to @xmath230 for any @xmath231 to get @xmath232 by the choice of @xmath207 it is easy to check that @xmath233 .
therefore , in view of ( [ 10.3.5 ] ) , we have @xmath234 for all @xmath231 . since @xmath235 for all @xmath47
, it follows from ( [ 10.6.2 ] ) that @xmath236 .
the proof is therefore complete .
+ by taking @xmath237 in ( [ 10.6.2 ] ) , the integer @xmath205 defined by rule [ rule1 ] can be estimated by @xmath238 where @xmath239 . in case @xmath49
is chosen such that @xmath240 for all @xmath47 for some constant @xmath241 , then @xmath242 it then follows from ( [ 10.6.1 ] ) that @xmath243 which implies that @xmath244 . therefore , with such a chosen @xmath49 , algorithm [ alg1 ] exhibits the fast convergence property . in order to use the results given in lemma [ lem1 ] and theorem [ thm1 ] to prove the convergence of the method ( [ method_noise ] ) , we need the following stability result . [ lem10.9.1 ]
let @xmath168 and @xmath82 be defined by ( [ method ] ) with @xmath54 chosen by ( [ 10.13.7 ] ) , and let @xmath217 and @xmath218 be defined by ( [ method_noise ] ) with @xmath245 chosen by ( [ 10.3.6 ] ) .
then for each fixed integer @xmath47 there hold @xmath246 * proof . *
we prove the result by induction on @xmath47 .
it is trivial when @xmath229 because @xmath247 and @xmath248 .
assume next that the result is true for some @xmath249 .
we will show that @xmath250 and @xmath251 as @xmath252 .
we consider two cases : _ case 1 : @xmath253 .
_ in this case we must have @xmath254 since otherwise @xmath255 therefore , by the induction hypothesis it is straightforward to see that @xmath256 as @xmath257 . according to the definition of @xmath211 and the induction hypothesis , we then obtain @xmath258 .
recall that @xmath259 we then obtain @xmath260 by the continuity of @xmath75 .
_ case 2 : @xmath261 .
_ in this case we have @xmath262 .
therefore @xmath263 consequently , by the induction hypothesis , we have @xmath264 by using again the continuity of @xmath75 , we obtain @xmath260 .
+ we are now in a position to give the main result concerning the regularization property of the method ( [ method_noise ] ) with noisy data when it is terminated by rule [ rule1 ] .
[ thm2 ] let @xmath6 $ ] be proper , lower semi - continuous and strong convex in the sense of ( [ sc ] ) .
let @xmath49 be a decreasing sequence of positive numbers and let @xmath245 be chosen by ( [ 10.3.6 ] ) with @xmath215 and @xmath99 .
let @xmath205 be the finite integer defined by rule [ rule1 ] .
then for the method ( [ method_noise ] ) there hold @xmath265 * proof . * due to the strong convexity of @xmath12 , it suffices to show that @xmath266 . by the subsequence - subsequence argument
, we may complete the proof by considering two cases .
assume first that @xmath267 is a sequence satisfying @xmath268 with @xmath269 such that @xmath270 as @xmath135 for some finite integer @xmath271 .
we may assume @xmath272 for all @xmath139 .
from the definition of @xmath273 we have @xmath274 by taking @xmath135 and using lemma [ lem10.9.1 ] , we can obtain @xmath275 . in view of the definition of @xmath168 and @xmath82
, this implies that @xmath276 and @xmath277 for all @xmath278 .
since theorem [ thm1 ] implies that @xmath279 as @xmath160 , we must have @xmath280 . moreover , by lemma [ lem10.9.1 ] , @xmath281 as @xmath135 . therefore , by the continuity of @xmath74 we can obtain @xmath282 assume next that @xmath267 is a sequence satisfying @xmath268 with @xmath269 such that @xmath283 as @xmath135 .
let @xmath47 be any fixed integer .
then @xmath284 for large @xmath139 .
it then follows from ( [ monotone ] ) in lemma [ lem1 ] that @xmath285 by using lemma [ lem10.9.1 ] and the continuity of @xmath74 we obtain @xmath286 since @xmath47 can be arbitrary and since theorem [ thm1 ] implies that @xmath287 as @xmath160 , we therefore have @xmath288 .
+ in certain applications , the solution of ( [ 1.1 ] ) may have some physical constraints .
thus , instead of ( [ 1.1 ] ) , we need to consider the constrained problem @xmath289 where @xmath290 is a closed convex subset in @xmath2 .
correspondingly , ( [ method_noise ] ) can be modified into the form @xmath291 which can be analyzed by the above framework by introducing @xmath292 , where @xmath293 denotes the indicator function of @xmath290 , i.e. @xmath294 when @xmath207 is chosen by ( [ 10.3.6 ] ) and ( [ 10.16.1 ] ) is terminated by rule [ rule1 ] , we still have @xmath295 and @xmath296 as @xmath257
. however , @xmath297 may not converge to @xmath298 because @xmath299 is not necessarily in @xmath300 . in order to implement algorithm [ alg1 ] , a key ingredient is to solve the minimization problem @xmath301 for any given @xmath302 .
for some choices of @xmath12 , this minimization problem can be efficiently solved numerically . when @xmath303 , where @xmath304 is a bounded lipschitz domain in euclidean space , there are at least two important choices of @xmath12 that are crucial for sparsity recovery and discontinuity detection .
the first one is @xmath305 with @xmath306 , the minimizer of ( [ eq4.16.1 ] ) can be given explicitly by @xmath307 the second one is @xmath308 with @xmath306 , where @xmath309 denotes the total variation of @xmath64 , i.e. @xmath310 then the minimization problem ( [ eq4.16.1 ] ) can be equivalently formulated as @xmath311 which is the total variation denoising problem ( @xcite ) .
although there is no explicit formula for the minimizer of ( [ eq4.16.1 ] ) , there are many efficient numerical solvers developed in recent years , see @xcite . for the numerical simulations involving total variation presented in section 3
, we always use the denoising algorithm ` fista ` from @xcite . indeed , when solving ( [ eq4.16.1 ] ) with @xmath12 given by ( [ eq : tv ] ) , `
fista ` is used to solve its dual problem whose solution determines the solution of the primal problem ( [ eq4.16.1 ] ) directly ; one may refer to the algorithm on page 2462 in @xcite and its monotone version after acceptance of this paper , we found that our method can be significantly accelerated if we use pdhg ( an application of uzawa algorithm ) to solve the tv denoising problem .
the matlab code of pdhg can be found at http://pages.cs.wisc.edu/@xmath312swright/gpureconstruction/ .
another key ingredient in implementing algorithm [ alg1 ] is to determine @xmath313 for @xmath314 , where @xmath315 .
this amounts to solving the linear equation @xmath316 for which many efficient solvers from numerical linear algebra can be applied .
when @xmath21 has special structure , this equation can even be solved very fast .
for instance , if @xmath21 is a convolution operator in @xmath317 , say @xmath318 with the kernel @xmath139 decaying sufficiently fast at infinity , then @xmath319 can be determined as @xmath320 where @xmath321 and @xmath322 denote the fourier transform and the inverse fourier transform respectively .
therefore @xmath319 can be calculated efficiently by the fast fourier transform .
our method can be extended for solving nonlinear inverse problems in hilbert spaces that can be formulated as the equation @xmath323 where @xmath324 is a nonlinear continuous operator between two hilbert spaces @xmath2 and @xmath3 with closed convex domain @xmath325 .
we assume that for each @xmath326 there is a bounded linear operator @xmath327 such that @xmath328 in case @xmath329 is frchet differentiable at @xmath64 , @xmath330 is exactly the frchet derivative of @xmath329 at that point . in order to find the solution of ( [ n1.1 ] ) with special feature , as before we introduce a penalty function @xmath6 $ ] which is proper , convex and lower semi - continuous . let @xmath16 be the only available noisy data satisfying @xmath331 with a small known noise level @xmath194
. then an obvious extension of algorithm [ alg1 ] for solving ( [ n1.1 ] ) takes the following form .
[ alg2 ] 1 .
take @xmath200 , @xmath208 , @xmath99 and a decreasing sequence @xmath49 of positive numbers ; 2 .
take @xmath78 and define @xmath332 as an initial guess ; 3 . for each @xmath210 define @xmath333 where @xmath334 4 .
let @xmath205 be the first integer such that @xmath335 and use @xmath336 to approximate the solution of ( [ n1.1 ] ) .
we remark that when @xmath337 , algorithm [ alg2 ] reduces to a method which is similar to the regularized levenberg - marquardt method in @xcite for which convergence is proved under certain conditions on @xmath329 . for general convex penalty function @xmath12 , however , we do not have convergence theory on algorithm [ alg2 ] yet .
nevertheless , we will give numerical simulations to indicate that it indeed performs very well .
in this section we will provide various numerical simulations on our method .
the choice of the sequence @xmath49 plays a crucial role for the performance : if @xmath49 decays faster , only fewer iterations are required but the reconstruction result is less accurate ; on the other hand , if @xmath49 decays slower , more iterations are required but the reconstruction result is more accurate . in order to solve this dilemma
, we choose fast decaying @xmath49 at the beginning , and then choose slow decaying @xmath49 when the method tends to stop .
more precisely , we choose @xmath49 according to the following rule .
[ alpha ] let @xmath338 and @xmath339 be preassigned numbers . we take some number @xmath340 and for @xmath249
define @xmath341 if @xmath342 we set @xmath343 ; otherwise we set @xmath344 .
all the computation results in this section are based on @xmath49 chosen by this rule with @xmath345 , @xmath346 and @xmath347 .
our tests were done by using matlab r2012a on an lenovo laptop with intel(r ) core(tm ) i5 cpu 2.30 ghz and 6 gb memory .
we first consider the integral equation of the form @xmath348,\ ] ] where @xmath349 it is easy to see , that @xmath21 is a compact linear operator from @xmath350 $ ] to @xmath350 $ ] . our goal is to find the solution of ( [ int ] ) using noisy data @xmath16 satisfying @xmath351}=\d$ ] for some specified noise level @xmath352 . in our numerical simulations , we divide @xmath353 $ ] into @xmath354 subintervals of equal length and approximate any integrals by the trapezoidal rule .
in figure [ fig1 ] we report the numerical performance of algorithm [ alg1 ] .
the sequence @xmath49 is selected by rule [ alpha ] with @xmath355 , @xmath356 , @xmath357 and @xmath358 .
the first row gives the reconstruction results using noisy data with various noise levels when the sought solution is sparse ; we use the penalty function @xmath12 given in ( [ eq : l1 ] ) with @xmath359 .
the second row reports the reconstruction results for various noise levels when the sought solution is piecewise constant ; we use the penalty function @xmath12 given in ( [ eq : tv ] ) with @xmath360 .
when the 1d tv - denoising algorithm ` fista ` in @xcite is used to solve the minimization problems associated with this @xmath12 , it is terminated as long as the number of iterations exceeds @xmath361 or the error between two successive iterates is smaller than @xmath362 . during these computations ,
we use @xmath363 and the parameters @xmath364 , @xmath365 and @xmath366 in algorithm [ alg1 ] .
the computational times for the first row are @xmath367 , @xmath368 and @xmath369 seconds respectively , and the computation times for the second row are @xmath370 , @xmath371 and @xmath372 seconds respectively .
this shows that algorithm [ alg1 ] indeed is a fast method with the capability of capturing special features of solutions .
let @xmath373\times [ 0,1]$ ] .
we consider the problem of determining the source term @xmath374 in the poisson equation @xmath375 from an @xmath376 measurement @xmath377 of @xmath378 with @xmath379 .
this problem takes the form ( [ 1.1 ] ) if we define @xmath380 , where @xmath381 is an isomorphism .
the information on @xmath21 can be obtained by solving the equation . in order to solve the poisson equation numerically ,
we take @xmath382 grid points @xmath383 on @xmath304 , and write @xmath384 for @xmath385 and @xmath386 for @xmath387 . by the finite difference representation of @xmath388 , the poisson equation has the discrete form @xmath389 where @xmath390 . since @xmath391 on @xmath392 , the discrete sine transform can be used to solve ( [ 10.29.1 ] ) .
consequently @xmath384 can be determined by the inverse discrete sine transform ( @xcite ) @xmath393 for @xmath394 , where @xmath395 and @xmath396 is determined by the discrete sine transform @xmath397 let @xmath398 .
then @xmath399 can be determined by solving the equation @xmath400 . when applying algorithm [ alg1 ] , we need to determine @xmath401 for various @xmath314 and vectors @xmath402 .
this can be computed as @xmath403 where , for any vector @xmath404 , @xmath405 and @xmath406 can be implemented by the fast sine and inverse sine transforms respectively , while @xmath407_{i , j}= { \bf w}_{i , j}/(\a + h^4/(4 - 2\cos(ih\pi)-2\cos(jh\pi))^2)\ ] ] therefore @xmath408 can be computed efficiently .
we apply algorithm [ alg1 ] to reconstruct the source term which is assumed to be piecewise constant . in our computation
we use a noisy data with noise level @xmath409 .
the left plot in figure [ fig4 ] is the exact solution . the right plot in figure [ fig4 ]
is the reconstruction result by algorithm [ alg1 ] using initial guess @xmath410 and the penalty function @xmath411 where @xmath412 is the frobenius norm of @xmath399 and @xmath413 denotes the discrete isotropic tv defined by ( @xcite ) @xmath414 in each step of algorithm [ alg1 ] , the minimization problem associate with @xmath12 is solved by performing 400 iterations of the 2d tv - denoising algorithm ` fista ` in @xcite . in our computation , we use @xmath415 , and for those parameters in algorithm [ alg1 ] , we take @xmath364 , @xmath416 , @xmath417 and @xmath418 . when using rule [ alpha ] to choose @xmath49 we take @xmath419 , @xmath420 , @xmath421 and @xmath422 .
the reconstruction result indicates that our method succeeds in capturing the feature of the solution .
moreover , the computation terminates after @xmath423 iterations and takes 11.7092 seconds . blurring in images
can arise from many sources , such as limitations of the optical system , camera and object motion , astigmatism , and environmental effects ( @xcite ) .
image deblurring is the process of making a blurry image clearer to better represent the true scene .
we consider grayscale digital images which can be represented as rectangular matrices of size @xmath424 .
let @xmath425 and @xmath426 denote the true image and the blurred image respectively .
the blurring process can be described by an operator @xmath427 such that @xmath428 .
we consider the case that the model is shift - invariant and @xmath429 is linear . by stacking the columns of @xmath425 and @xmath426
we can get two long column vectors @xmath430 and @xmath431 of length @xmath432 . then there is a large matrix @xmath433 such that @xmath434 . considering the appearance of unavoidable random noise
, one in fact has @xmath435 where @xmath436 denotes the noise . the blurring matrix @xmath437 is determined by the point spread function ( psf ) @xmath438the function that describes the blurring and the resulting image of the single bright pixel ( i.e. point source ) . throughout this subsection ,
periodic boundary conditions are assumed on all images .
then @xmath437 is a matrix which is block circulant with circulant blocks ; each block is built from @xmath438 .
it turns out that @xmath437 has the spectral decomposition @xmath439 where * f * is the two - dimensional unitary discrete fourier transform matrix and @xmath440 is the diagonal matrix whose diagonal entries are eigenvalues of @xmath437 .
the diagonal matrix @xmath440 is easily determined by the smaller matrix @xmath438 , and the action of @xmath441 and @xmath442 can be realized by ` fft ` and ` ifft ` .
therefore , for any @xmath443 and @xmath314 , @xmath444 is easily computable by the fast fourier transform . in the following we perform some numerical experiments by applying algorithm [ alg1 ] to deblur various corrupted images . in our simulations the exact data @xmath431
are contaminated by random noise vectors @xmath436 whose entries are normally distributed with zero mean .
we use @xmath445 to denote the relative noise level . when applying algorithm [ alg1 ] , we use @xmath446 and the following two convex functions @xmath447 for those parameters in the algorithm we take @xmath448 , @xmath449 and @xmath418 . in each step of the algorithm , the minimization problem associate with @xmath12 is solved by performing 200 iterations of the algorithm ` fista ` in @xcite . when using rule [ alpha ] to choose @xmath49 we take @xmath450 , @xmath420 , @xmath451 and @xmath452 .
in order to compare the quality of the restoration @xmath453 , we evaluate the peak signal - to - noise ratio ( psnr ) value defined by @xmath454 where @xmath455 denotes the maximum possible pixel value of the true image @xmath425 .
in figure [ fig5 ] we plot the restoration results of the shepp - logan phantom of size @xmath456 which is blurred by a @xmath457 gaussian psf with standard derivation @xmath458 and is contaminated by gaussian white noise with relative noise level @xmath459 .
the original and blurred images are plotted in ( a ) and ( b ) of figure [ fig5 ] respectively . in figure [ fig5 ] ( c )
we plot the restoration result by algorithm [ alg1 ] with @xmath460 . with such chosen @xmath12 ,
the method in algorithm [ alg1 ] reduces to the classical nonstationary iterated tikhonov regularization ( [ nsit ] ) which has the tendency to over - smooth solutions .
the plot clearly indicates this drawback because of the appearance of the ringing artifacts .
the corresponding psnr value is @xmath461 . in figure [ fig5 ] ( d )
we plot the restoration result by algorithm [ alg1 ] with @xmath462 . due to the appearance of the total variation term in @xmath463 ,
the artifacts are significantly removed .
in fact the corresponding psnr value is @xmath464 ; the computation terminates after @xmath465 iterations and takes @xmath466 seconds . in figure [ fig6 ]
we plot the restoration results of the @xmath467 cameraman image corrupted by a @xmath468 linear motion kernel generated by ` fspecial(motion,30,40 ) ` and a gaussian white noise with relative noise level @xmath469 .
the original and blurred images are plotted in ( a ) and ( b ) of figure [ fig6 ] respectively . in ( c ) and ( d ) of figure [ fig6 ] we plot the restoration results by algorithm [ alg1 ] with @xmath460 and @xmath470 respectively .
the plot in ( c ) contains artifacts that degrade the visuality , the plot in ( d ) removes the artifacts significantly .
in fact the psnr values corresponding to ( c ) and ( d ) are @xmath471 and @xmath472 respectively . the computation for ( d ) terminates after @xmath473 iterations and takes @xmath474 seconds .
we finally present some numerical simulations for nonlinear inverse problems by solving the autoconvolution equation @xmath475 defined on the interval @xmath476 $ ] .
the properties of the autoconvolution operator @xmath477(t ) : = \int_0^t x(t - s ) x(s ) ds$ ] have been discussed in @xcite . in particular , as an operator from @xmath350 $ ] to @xmath350 $ ] , @xmath329 is frchet differentiable ; its frchet derivative and the adjoint are given respectively by @xmath478(t ) = 2 \int_0^t x(t - s ) v(s ) ds , \quad v \in l^2[0,1 ] , \\ \left[f'(x)^ * w\right](s ) = 2 \int_s^1 w(t ) x(t - s ) dt , \quad
w\in l^2[0,1].\end{aligned}\ ] ] we assume that ( [ autoconv ] ) has a piecewise constant solution and use a noisy data @xmath16 satisfying @xmath479}=\d$ ] to reconstruct the solution . in figure [ fig7 ]
we report the reconstruction results by algorithm [ alg2 ] using @xmath480 and the @xmath12 given in ( [ eq : tv ] ) with @xmath481 .
all integrals involved are approximated by the trapezoidal rule by dividing @xmath353 $ ] into @xmath354 subintervals of equal length . for those parameters involved in the algorithm ,
we take @xmath482 , @xmath483 and @xmath484 .
we also take the constant function @xmath485 as an initial guess .
the sequence @xmath49 is selected by rule [ alpha ] with @xmath21 replaced by @xmath486 in which @xmath487 , @xmath420 , @xmath357 and @xmath488 . when the 1d - denoising algorithm ` fista ` in @xcite is used to solve the minimization problems associated with @xmath12 , it is terminated as long as the number of iterations exceeds @xmath489 or the error between two successive iterates is smaller than @xmath490 .
we indicate in figure [ fig7 ] the number of iterations and the computational time for various noise levels @xmath491 ; the results show that algorithm [ alg2 ] is indeed a fast method for this problem .
we proposed a nonstationary iterated method with convex penalty term for solving inverse problems in hilbert spaces .
the main feature of our method is its splitting character , i.e. each iteration consists of two steps : the first step involves only the operator from the underlying problem so that the hilbert space structure can be exploited , while the second step involves merely the penalty term so that only a relatively simple strong convex optimization problem needs to be solved .
this feature makes the computation much efficient . when the underlying problem is linear , we proved the convergence of our method in the case of exact data ; in case only noisy data are available , we introduced a stopping rule to terminate the iteration and proved the regularization property of the method .
we reported various numerical results which indicate the good performance of our method .
q. jin is partially supported by the decra grant de120101707 of australian research council and x. lu is partially supported by national science foundation of china ( no . 11101316 and no .
91230108 ) .
999 h. gfrerer , _ an a posteriori parameter choice for ordinary and iterated tikhonov regularization of ill - posed problems leading to optimal convergence rates _ , math . comp . , 49(180 ) : 507522 , s5s12 , 1987 . | in this paper we consider the computation of approximate solutions for inverse problems in hilbert spaces . in order to capture the special feature of solutions ,
non - smooth convex functions are introduced as penalty terms . by exploiting the hilbert space structure of the underlying problems ,
we propose a fast iterative regularization method which reduces to the classical nonstationary iterated tikhonov regularization when the penalty term is chosen to be the square of norm .
each iteration of the method consists of two steps : the first step involves only the operator from the problem while the second step involves only the penalty term .
this splitting character has the advantage of making the computation efficient . in case
the data is corrupted by noise , a stopping rule is proposed to terminate the method and the corresponding regularization property is established . finally , we test the performance of the method by reporting various numerical simulations , including the image deblurring , the determination of source term in poisson equation , and the de - autoconvolution problem . |
the present work is inspired by the work of barberis @xcite where invariant hypercomplex structures @xmath0 on 4-dimensional real lie groups are classified . in that case
the corresponding metric is positive definite and hermitian with respect to the triplet of complex structures of @xmath0 .
our main goal is to classify 4-dimensional real lie algebras which admit hypercomplex structures with hermitian and norden metrics .
we equip a hypercomplex structure @xmath0 with a metric structure , generated by a pseudo - riemannian metric @xmath1 of neutral signature ( see @xcite ) . in our case the one ( resp . , the other two ) of the almost complex structures of @xmath0 acts as an isometry ( resp . , act as anti - isometries ) with respect to @xmath1 in each tangent fibre .
thus , there exist three ( 0,2)-tensors associated by @xmath0 except the metric @xmath1 a khler form and two metrics of the same type .
the metric @xmath1 is hermitian with respect to the one almost complex structure of @xmath0 and @xmath1 is a norden ( i.e. an anti - hermitian ) metric regarding the other two almost complex structures of @xmath0 .
for this reason we call the derived almost hypercomplex structure an _ almost hypercomplex structure with hermitian - norden - metrics _ or briefly _ almost hypercomplex hn - metric structure_. let us remark that in @xcite and @xcite are classified the invariant complex structures on 4-dimensional solvable simply - connected real lie groups where the dimension of commutators is less than three and equal three , respectively .
a hypercomplex structure is called _ abelian _ ( @xcite ) if @xmath2 = [ x , y]$ ] , for all @xmath3 ( @xmath4 ) .
abelian hypercomplex structures are considered in @xcite , @xcite and they can only occur on solvable lie algebras ( @xcite ) .
it is clear that the condition @xmath5-j[jx , y]-j[x , jy]-[x , y]=0 $ ] can be rewritten as @xmath6-[x , y ] = j([jx , y]+[x , jy])$ ] for all @xmath3 .
thus , abelian complex structures and therefore abelian hypercomplex structure are integrable .
if the three almost complex structures of @xmath0 are parallel with respect to the levi - civita connection @xmath7 of @xmath1 then such hypercomplex hn - manifolds of khler type we call _ hyper - khler hn - manifolds _ , which are flat according to @xcite . the paper is organized as follows . in sect .
[ sec1 ] we recall some facts about the almost hypercomplex hn - manifolds known from @xcite . in sect . [ sec2 ] we construct different types of hypercomplex structures on lie algebras following the barberis classification .
the basic problem of this work is the existence and the geometric characteristics of hypercomplex hn - structures on 4-dimensional lie algebras according to the barberis classification .
the main results of this paper is construction of the different types of the considered structures and their characterization .
let @xmath8 be a hypercomplex manifold , i.e. @xmath9 is a @xmath10-dimensional differentiable manifold and @xmath11 is a triple of complex structures on @xmath9 with the following properties for all cyclic permutations @xmath12 of @xmath13:@xmath14 where @xmath15 denotes the identity and moreover , it is valid @xmath16 for the nijenhuis tensors @xmath17 of @xmath18 given by @xmath19 -j_{\alpha}\left[j_{\alpha}\cdot,\cdot \right ] -j_{\alpha}\left[\cdot , j_{\alpha}\cdot \right ] -\left[\cdot,\cdot \right]\]]on @xmath20 . a hypercomplex structure on a lie group @xmath21 is said to be _ invariant _ if left translations by elements of @xmath21 are holomorphic with respect to @xmath22 for all @xmath23 .
obviously , if @xmath24 is the corresponding lie algebra of the lie group @xmath21 , a hypercomplex structure on @xmath24 induces an invariant hypercomplex structure on @xmath21 by left translations .
let @xmath21 be a simply connected 4-dimensional real lie group admitting an invariant hypercomplex structure .
a left invariant metric on @xmath21 is called invariant hyper - hermitian if it is hyper - hermitian with respect to some invariant hypercomplex structure on @xmath21 .
it is known that all such metrics on given @xmath21 are equivalent up to homotheties .
if @xmath24 denotes the lie algebra of @xmath21 then it is known the following [ thm - barb ] the only 4-dimensional lie algebras admitting an integrable hypercomplex structure are the following types : _ * ( hc1 ) * _ @xmath24 is abelian;@xmath25 _ * ( hc2 ) * _ @xmath26 ; @xmath25 _ * ( hc3 ) * _ @xmath27 ; _ * ( hc4 ) * _ @xmath24 is the solvable lie algebra corresponding to @xmath28 ; _ * ( hc5 ) * _ @xmath24 is the solvable lie algebra corresponding to @xmath29 , + where @xmath30 is the lie algebra of the lie groups @xmath31 and @xmath32 ; @xmath33 is the lie algebra of the affine motion group of @xmath34 the unique 4-dimensional lie algebra carrying an abelian hypercomplex structure ; @xmath28 is the real hyperbolic space ; @xmath29 is the complex hyperbolic space .
let @xmath1 be a neutral metric on @xmath8 with the properties @xmath35 where @xmath36 moreover , the associated ( khler ) 2-form @xmath37 and the associated neutral metrics @xmath38 and @xmath39 are determined by @xmath40 the structure tensors of a such manifold are the following three @xmath41-tensors @xmath42 where @xmath43 is the levi - civita connection generated by @xmath1 .
the corresponding lee 1-forms @xmath44 are defined by @xmath45for an arbitrary basis @xmath46 of @xmath47 , @xmath48 .
in @xcite we study the so - called _ hyper - khler manifolds with hn - metric structure _ ( or pseudo - hyper - khler manifolds ) , i.e. the almost hypercomplex hn - manifold in the class @xmath49 , where @xmath50 for all @xmath51 .
a sufficient condition @xmath52 be in @xmath49 is this manifold be of khler - type with respect to two of the three complex structures of @xmath0 @xcite .
as @xmath1 is an indefinite metric , there exist isotropic vectors @xmath53 on @xmath9 , i.e. @xmath54 , @xmath55 .
in @xcite we define the invariant square norm @xmath56 where @xmath46 is an arbitrary basis of the tangent space @xmath47 at an arbitrary point @xmath48 of @xmath47 .
we say that an almost hypercomplex hn - manifold is an _ isotropic hyper - khler hn - manifold _ if @xmath57 for each @xmath18 of @xmath0 .
clearly , if the manifold is a hyper - khler hn - manifold , then it is an isotropic hyper - khler hn - manifold .
the inverse statement does not hold .
let us note that according to the manifold @xmath58 is almost hermitian and the manifolds @xmath59 , @xmath60 , are almost complex manifolds with norden metric @xcite .
the basic classes of the mentioned two types of manifolds are given in @xcite and @xcite , respectively . the special class @xmath61 @xmath62 @xmath63 of the khler - type manifolds belongs to any other class within the corresponding classification . in the 4-dimensional case
the four basic classes of the almost hermitian manifolds are restricted to two : @xmath64 , the class of the almost khler manifolds and @xmath65 , the class of the hermitian manifolds with respect to @xmath66 .
they are determined for @xmath67 by : @xmath68 where @xmath69 is the cyclic sum by three arguments @xmath53 , @xmath70 , @xmath71 .
the basic classes of the almost norden manifolds ( i.e. , for @xmath72 or @xmath73 ) are determined for dimension @xmath74 as follows : @xmath75 it is known that the class of the complex manifolds with norden metric is @xmath76 for @xmath22 ( @xmath60 ) .
then the class of hypercomplex manifolds with hermitian - norden metrics is @xmath77
let @xmath78 be a basis of a 4-dimensional real lie algebra @xmath24 with center @xmath79 and derived lie algebra @xmath80 $ ] .
a standard hypercomplex structure on @xmath24 is defined as in @xcite : @xmath81 j_2e_1=e_3 , & j_2e_2=e_4 , & j_2e_3=-e_1 , & j_2e_4=-e_2 ; \\[4pt ] j_3e_1=-e_4 , & j_3e_2=e_3 , & j_3e_3=-e_2 , & j_3e_4=e_1.\\[4pt ] \end{array}\ ] ] let us introduced a pseudo - euclidian metric @xmath1 with neutral signature as follows @xmath82 where @xmath83 , @xmath84 .
this metric satisfies and .
then the metric @xmath1 generates an almost hypercomplex hn - metric structure on @xmath24 .
let @xmath24 be not solvable and let us determine it by @xmath85=e_3,\qquad [ e_4,e_3]=e_2,\qquad [ e_3,e_2]=e_4 . \ ] ] in this consideration the @xmath86-unit @xmath87 , i.e. @xmath88 , is orthogonal to @xmath89 with respect to @xmath1 .
then we compute covariant derivatives in the basis and the nontrivial ones are @xmath90 { \nabla}_{e_2 } e_4=\frac{3}{2}e_3,\qquad & { \nabla}_{e_3 } e_4=-\frac{1}{2}e_2,\qquad & { \nabla}_{e_4 } e_3=\frac{1}{2}e_2 . \end{array}\ ] ] by virtue of , and , we obtain components @xmath91 , @xmath92 , as follows : @xmath93 & = -(f_{1})_{413}=-(f_{1})_{424}=(f_{1})_{431}=(f_{1})_{442}=\frac{1}{2};\\[4pt ] \end{aligned}\\ & \begin{aligned } ( f_{2})_{214}&=-(f_{2})_{223}=-(f_{2})_{232}=(f_{2})_{241}=\frac{3}{2},\quad & ( f_{2})_{322}=-1 , \\[4pt ] ( f_{2})_{412}&=(f_{2})_{421}=(f_{2})_{434}=(f_{2})_{443}=\frac{1}{2 } , \quad & ( f_{2})_{344}=-1;\\[4pt ] ( f_{3})_{213}&=(f_{3})_{224}=(f_{3})_{231}=(f_{3})_{242}=\frac{3}{2},\quad & ( f_{3})_{422}=1 , \\[4pt ] ( f_{3})_{312}&=(f_{3})_{321}=-(f_{3})_{334}=-(f_{3})_{343}=\frac{1}{2 } , \quad & ( f_{3})_{433}=1 . \end{aligned } \end{aligned}\ ] ] the only non - zero components @xmath94 , @xmath95 , of the corresponding lee forms are @xmath96 using the results in , and the classification conditions , , we obtain [ prop - hc2 ] the hypercomplex manifold with hermitian - norden metrics on a 4-dimensional lie algebra , determined by , belongs to the largest class of the considered manifolds , i.e. @xmath97 , as well as this manifold does not belong to neither @xmath98 nor @xmath99 for @xmath100 and @xmath101 .
the other possibility is the signature of @xmath1 on @xmath102 to be @xmath103 , e.g. @xmath104 , where @xmath105 . by similar computations we establish the same class in the statement of proposition [ prop - hc2 ] .
let us determine @xmath24 by @xmath106=[e_1,e_4]=e_2,\qquad [ e_2,e_1]=[e_4,e_3]=e_4.\ ] ] then we compute covariant derivatives and the nontrivial ones are @xmath107 { \nabla}_{e_2 } e_3=-{\nabla}_{e_4 } e_1=e_2,\qquad & { \nabla}_{e_2 } e_4={\nabla}_{e_4 } e_2=e_1 . \end{array}\ ] ] by virtue of , and , we obtain that @xmath108 and the other components @xmath109 , @xmath60 , are as follows @xmath110 & = -(f_{2})_{414}=(f_{2})_{423}=(f_{2})_{432}=-(f_{2})_{441}=2 ; \\[4pt ] ( f_{3})_{211}&=-(f_{3})_{222}=-(f_{3})_{233}=(f_{3})_{244}= \\[4pt ] & = ( f_{3})_{413}=(f_{3})_{424}=(f_{3})_{431}=(f_{3})_{442}=-2
. \end{aligned}\ ] ] the only non - zero components of the corresponding lee forms are @xmath111 using that @xmath108 , the results in , and the classification conditions , , we obtain [ prop - hc3 ] the hypercomplex manifold with hermitian - norden metrics on a 4-dimensional lie algebra , determined by , belongs to the subclass of the khler manifold with respect to @xmath66 of the largest class of the considered manifolds , i.e. @xmath112 as well as this manifold does not belong to neither @xmath98 nor @xmath99 for @xmath100 and @xmath101 . let us determine @xmath24 by @xmath113=[e_4,e_2]=e_1,\qquad [ e_1,e_4]=[e_2,e_3]=e_2.\ ] ] then we compute covariant derivatives and the nontrivial ones are @xmath114 { \nabla}_{e_1 } e_3={\nabla}_{e_4 } e_2=e_1 . \end{array}\ ] ] by virtue of , and , we obtain the following components of @xmath115 : @xmath116 & = ( f_{1})_{213}=(f_{1})_{224}=-(f_{1})_{231}=-(f_{1})_{242}=-1 ; \\[4pt ] ( f_{2})_{111}&=(f_{2})_{133}=2,\\[4pt ] ( f_{2})_{212}&=(f_{2})_{221}=(f_{2})_{234}=(f_{2})_{243 } \\[4pt ] & = -(f_{2})_{414}=(f_{2})_{423}=(f_{2})_{432}=-(f_{2})_{441}=1 ; \\[4pt ] ( f_{3})_{222}&=(f_{3})_{233}=2,\\[4pt ] -(f_{3})_{112}&=-(f_{3})_{121}=(f_{3})_{134}=(f_{3})_{143 } \\[4pt ] & = ( f_{3})_{413}=(f_{3})_{424}=(f_{3})_{431}=(f_{3})_{442}=-1 . \end{aligned}\ ] ] the only non - zero components of the corresponding lee forms are @xmath117 using the results in , and the classification conditions , , we obtain that the considered manifold belongs to the class @xmath118 .
remark that , according to @xcite , necessary and sufficient conditions a 4-dimensional almost hypercomplex hn - manifold to be in the class @xmath119 are : @xmath120 these conditions are satisfied bearing in mind .
let us consider the class @xmath121 , which is the class of the ( locally ) conformally equivalent @xmath49-manifolds , where a conformal transformation of the metric is given by @xmath122 for a differential function @xmath123 on the manifold .
[ prop - hc3b ] the hypercomplex manifold with hermitian - norden metrics on a 4-dimensional lie algebra , determined by , belongs to the class @xmath124 of the ( locally ) conformally equivalent @xmath49-manifolds .
firstly , we fix @xmath125 , for which @xmath88 , as an element orthogonal to @xmath89 with respect to @xmath1 . therefore @xmath24 is determined by @xmath126=e_2,\qquad [ e_1,e_3]=e_3,\qquad [ e_1,e_4]=e_4.\ ] ] then we compute covariant derivatives and the nontrivial ones are @xmath127 { \nabla}_{e_2 } e_2=-{\nabla}_{e_3 } e_3=-{\nabla}_{e_4 } e_4=e_1 .
\end{array}\ ] ] by similar computation as in the previous cases , the components @xmath109 , @xmath51 , are as follows : @xmath128 & = -(f_{1})_{413}=-(f_{1})_{424}=(f_{1})_{431}=(f_{1})_{442}=1 ; \\[4pt ] ( f_{2})_{311}&=(f_{2})_{333}=-2,\\[4pt ] ( f_{2})_{214}&=-(f_{2})_{223}=-(f_{2})_{232}=(f_{2})_{241 } \\[4pt ] & = ( f_{2})_{412}=(f_{2})_{421}=(f_{2})_{434}=(f_{2})_{443}=-1 ; \\[4pt ] ( f_{3})_{411}&=(f_{3})_{444}=2,\\[4pt ] ( f_{3})_{213}&=(f_{3})_{224}=(f_{3})_{231}=(f_{3})_{242 } \\[4pt ] & = ( f_{3})_{312}=(f_{3})_{321}=-(f_{3})_{334}=-(f_{3})_{343}=-1 .
\end{aligned}\ ] ] the only non - zero components of the corresponding lee forms are @xmath129 the results in , and the classification conditions , imply [ prop - hc4 ] the hypercomplex manifold with hermitian - norden metrics on a 4-dimensional lie algebra , determined by , belongs to the largest class of the considered manifolds , i.e. @xmath97 , as well as this manifold does not belong to neither @xmath98 nor @xmath99 for @xmath100 and @xmath101 . secondly , we choose @xmath130 , for which @xmath131 , as an element orthogonal to @xmath89 with respect to @xmath1 . therefore , in this case @xmath24 is determined by @xmath132=e_1,\qquad [ e_4,e_2]=e_2,\qquad [ e_4,e_3]=e_3.\ ] ] therefore , the nontrivial covariant derivatives are @xmath133 { \nabla}_{e_1 } e_4=-e_1 , \qquad { \nabla}_{e_2 } e_4=-e_2 , \qquad { \nabla}_{e_3 } e_4=-e_3 . \end{array}\ ] ] in a similar way we obtain : @xmath134 & = -(f_{1})_{214}=(f_{1})_{223}=-(f_{1})_{232}=(f_{1})_{241}=-1 ; \\[4pt ] ( f_{2})_{222}&=(f_{2})_{244}=-2,\\[4pt ] ( f_{2})_{112}&=(f_{2})_{121}=(f_{2})_{134}=(f_{2})_{143 } \\[4pt ] & = ( f_{2})_{314}=-(f_{2})_{323}=-(f_{2})_{332}=(f_{2})_{341}=-1 ; \\[4pt ] ( f_{3})_{111}&=(f_{3})_{144}=2,\\[4pt ] -(f_{3})_{212}&=-(f_{3})_{221}=(f_{3})_{234}=(f_{3})_{243 } \\[4pt ] & = ( f_{3})_{313}=(f_{3})_{324}=(f_{3})_{331}=(f_{3})_{342}=-1 .
\end{aligned}\ ] ] the only non - zero components of the corresponding lee forms are @xmath135 [ prop - hc4b ] the hypercomplex manifold with hermitian - norden metrics on a 4-dimensional lie algebra , determined by , belongs to the class @xmath124 of the ( locally ) conformally equivalent @xmath49-manifolds .
firstly , we fix @xmath125 , for which @xmath88 , as an element orthogonal to @xmath89 with respect to @xmath1
. then @xmath24 is determined by @xmath136=e_2,\quad [ e_1,e_3]=\frac{1}{2}e_3,\quad [ e_1,e_4]=\frac{1}{2}e_4,\quad [ e_3,e_4]=\frac{1}{2}e_2.\ ] ] then we compute covariant derivatives and the nontrivial ones are @xmath137 -{\nabla}_{e_2 } e_1=4{\nabla}_{e_3 } e_4=-4{\nabla}_{e_4 } e_3=e_2 , & \\[4pt ] -4{\nabla}_{e_2 } e_4=-2{\nabla}_{e_3 } e_1=-4{\nabla}_{e_4 } e_2=e_3 , & \\[4pt ] 4{\nabla}_{e_2 } e_3=4{\nabla}_{e_3 } e_2=-2{\nabla}_{e_4 } e_1=e_4 . &
\end{aligned}\ ] ] analogously of the previous cases we obtain the non - zero components @xmath109 , @xmath51 , as follows : @xmath138 & = -(f_{1})_{413}=-(f_{1})_{424}=(f_{1})_{431}=(f_{1})_{442}=\frac{1}{4 } ; \\[4pt ] ( f_{2})_{214}&=-(f_{2})_{223}=-(f_{2})_{232}=(f_{2})_{241}=-\frac{5}{4 } , \\[4pt ] ( f_{2})_{311}&=-2(f_{2})_{322}=(f_{2})_{333}=-2(f_{2})_{344}=-1,\\[4pt ] ( f_{2})_{412}&=(f_{2})_{421}=(f_{2})_{434}=(f_{2})_{443}=-\frac{3}{4 } ; \\[4pt ] ( f_{3})_{213}&=(f_{3})_{224}=(f_{3})_{231}=(f_{3})_{242}=-\frac{5}{4 } , \\[4pt ] ( f_{3})_{312}&=(f_{3})_{321}=-(f_{3})_{334}=-(f_{3})_{343}=-\frac{3}{4 } , \\[4pt ] ( f_{3})_{411}&=-2(f_{3})_{422}=-2(f_{3})_{433}=(f_{3})_{444}=1 . \end{aligned}\ ] ] the only non - zero components of the corresponding lee forms are @xmath139 [ prop - hc5a ] the hypercomplex manifold with hermitian - norden metrics on a 4-dimensional lie algebra , determined by , belongs to the largest class of the considered manifolds , i.e. @xmath97 , as well as this manifold does not belong to neither @xmath98 nor @xmath99 for @xmath100 and @xmath101 .
the other possibility is to choose @xmath130 , for which @xmath131 , as an element orthogonal to @xmath89 with respect to @xmath1 .
we rearrange the basis in and then @xmath24 is determined by @xmath140=-\frac{1}{2}e_3,\quad [ e_1,e_4]=-\frac{1}{2}e_1 , \quad[e_2,e_4]=-\frac{1}{2}e_2,\quad [ e_3,e_4]=-e_3 . \ ] ] gribachev , k. , manev , m. : _ almost hypercomplex pseudo - hermitian manifolds and a 4-dimensional lie group with such structure_. j. geom .
* 88 * ( 1 - 2 ) , 4152 ( 2008 ) gribachev , k. , manev , m. , dimiev , s. : _ on the almost hypercomplex pseudo - hermitian manifolds_. in : dimiev , s. , sekigawa , k. ( eds . ) trends of complex analysis , differential geometry and mathematical physics , pp .
world sci .
publ . , singapore ( 2003 ) | integrable hypercomplex structures with hermitian and norden metrics on lie groups of dimension 4 are considered .
the corresponding five types of invariant hypercomplex structures with hyper - hermitian metric , studied by m.l .
barberis , are constructed here .
the different cases regarding the signature of the basic pseudo - riemannian metric are considered . |
a major challenge in neuroscience is to understand how the neural activities are propagated through different brain regions , since many cognitive tasks are believed to involve this process ( vogels and abbott , 2005 ) .
the feedforward neuronal network is the most used model in investigating this issue , because it is simple enough yet can explain propagation activities observed in experiments .
in recent years , two different modes of neural activity propagation have been intensively studied .
it has been found that both the synchronous spike packet ( _ synfire _ ) , and the _ firing rate _ , can be transmitted across deeply layered networks ( abeles 1991 ; aertsen et al .
1996 ; diesmann et al . 1999 ; diesmann et al . 2001 ; c@xmath0teau and fukai 2001 ; gewaltig et al .
2001 ; tetzlaff et al .
2002 ; tetzlaff et al .
2003 ; van rossum et al . 2002
; vogels and abbott 2005 ; wang et al . 2006 ; aviel et al . 2003 ; kumar et al . 2008 ; kumar et al . 2010 ; shinozaki et al . 2007 ; shinozaki et al .
although these two propagation modes are quite different , the previous results demonstrated that a single network with different system parameters can support stable and robust signal propagation in both of the two modes , for example , they can be bridged by the background noise and synaptic strength ( van rossum et al . 2002 ; masuda and aihara 2002 ; masuda and aihara 2003 ) .
neurons and synapses are fundamental components of the brain . by sensing outside signals , neurons continually fire discrete electrical signals known as action potentials or so - called spikes , and then transmit them to postsynaptic neurons through synapses ( dayan and abbott 2001 ) .
the spike generating mechanism of cortical neurons is generally highly reliable .
however , many studies have shown that the communication between neurons is , by contrast , more or less unreliable ( abeles 1991 ; raastad et al .
1992 ; smetters and zador 1996 ) .
theoretically , the synaptic unreliability can be explained by the phenomenon of probabilistic transmitter release ( branco and staras 2009 ; katz 1966 ; katz 1969 ; trommershuser et al .
1999 ) , i.e. , synapses release neurotransmitter in a stochastic fashion , which has been confirmed by well - designed biological experiments ( allen and stevens 1994 ) . in most cases , the transmission failure rate at a given synapse tends to exceed the fraction of successful transmission ( rosenmund et al .
1993 ; stevens and wang 1995 ) . in some special cases
, the synaptic transmission failure rate can be as high as 0.9 or even higher ( allen and stevens 1994 ) .
further computational studies have revealed that the unreliability of synaptic transmission might be a part of information processing of the brain and possibly has functional roles in neural computation .
for instance , it has been reported that the unreliable synapses provide a useful mechanism for reliable analog computation in space - rate coding ( maass and natschl@xmath1ger 2000 ) ; and it has been found that suitable synaptic successful transmission probability can improve the information transmission efficiency of synapses ( goldman 2004 ) and can filter the redundancy information by removing autocorrelations in spike trains ( goldman et al .
furthermore , it has also been demonstrated that unreliable synapses largely influence both the emergence and dynamical behaviors of clusters in an all - to - all pulse - coupled neuronal network , and can make the whole network relax to clusters of identical size ( friedrich and kinzel 2009 ) .
although the signal propagation in multilayered feedforward neuronal networks has been extensively studied , to the best of our knowledge the effects of unreliable synapses on the propagation of neural activity have not been widely discussed and the relevant questions still remain unclear ( but see the footnote ) . in this paper , we address these questions and provide insights by computational modeling . for this purpose
, we examine both the synfire propagation and firing rate propagation in feedforward neuronal networks .
we mainly investigate the signal propagation in feedforward neuronal networks composed of purely excitatory neurons connected with unreliable synapses in an all - to - all coupling fashion ( abbr .
ure feedforward neuronal network ) in this work .
we also compare our results with the corresponding feedforward neuronal networks ( we will clarify the meaning of `` corresponding '' later ) composed of purely excitatory neurons connected with reliable synapses in a random coupling fashion ( abbr .
rre feedforward neuronal network ) .
moreover , we study feedforward neuronal networks consisting of both excitatory and inhibitory neurons connected with unreliable synapses in an all - to - all coupling fashion ( abbr .
urei feedforward neuronal network ) .
the rest of this paper is organized as follows .
the network architecture , neuron model , and synapse model used in this paper are described in sec .
[ sec:2 ] . besides these
, the measures to evaluate the performance of synfire propagation and firing rate propagation , as well as the numerical simulation method are also introduced in this section .
the main results of the present work are presented in sec .
[ sec:3 ] .
finally , a detailed conclusion and discussion of our work are given in sec .
[ sec:4 ] .
in this subsection , we introduce the network topology used in this paper . here
we only describe how to construct the ure feedforward neuronal network . the methods about how to build
the corresponding rre feedforward neuronal network and the urei feedforward neuronal network will be briefly given in secs .
[ sec:3d ] and [ sec:3e ] , respectively .
the architecture of the ure feedforward neuronal network is schematically shown in figure [ fig:1 ] .
the network totally contains @xmath2 layers , and each layer is composed of @xmath3 excitatory neurons . since neurons in the first layer are responsible for receiving and encoding the external input signal , we therefore call this layer sensory layer and neurons in this layer are called sensory neurons . in contrast , the function of neurons in the other layers is to propagate neural activities .
based on this reason , we call these layers transmission layers and the corresponding neurons cortical neurons . because the considered neuronal network is purely feedforward , there is no feedback connection from neurons in downstream layers to neurons in upstream layers , and there is also no connection among neurons within the same layer . for simplicity , we call the @xmath4-th neuron in the @xmath5-th layer neuron @xmath6 in the following . network architecture of the ure feedforward neuronal network .
the network totally contains 10 layers .
the first layer is the sensory layer and the others are the transmission layers .
each layer consists of 100 excitatory neurons . for clarity ,
only 6 neurons are shown in each layer.,width=566 ] we now introduce the neuron model used in the present work .
each cortical neuron is modeled by using the integrate - and - fire ( if ) model ( nordlie et al .
2009 ) , which is a minimal spiking neuron model to mimic the action potential firing dynamics of biological neurons .
the subthreshold dynamics of a single if neuron obeys the following differential equation : @xmath7 with the total input current @xmath8 here @xmath9 and @xmath10 , @xmath11 represents the membrane potential of neuron @xmath6 , @xmath12 ms is the membrane time constant , @xmath13 mv is the resting membrane potential , @xmath14 m@xmath15 denotes the membrane resistance , and @xmath16 is the total synaptic current .
the noise current @xmath17 represents the external or intrinsic fluctuations of the neuron , where @xmath18 is a gaussian white noise with zero mean and unit variance , and @xmath19 is referred to as the noise intensity of the cortical neurons . in this work , a deterministic threshold - reset mechanism
is implemented for spike generation .
whenever the membrane potential of a neuron reaches a fixed threshold at @xmath20 mv , the neuron fires a spike , and then the membrane potential is reset according to the resting potential , where it remains clamped for a 5-ms refractory period .
on the other hand , we use different models to simulate the sensory neurons depending on different tasks . to study the synfire propagation
, we assume that each sensory neuron is a simple spike generator , and control their firing behaviors by ourselves .
while studying the firing rate propagation , the sensory neuron is modeled by using the if neuron model with the same expression ( see eq .
( [ eq:1 ] ) ) and the same parameter settings as those for cortical neurons . for each sensory neuron ,
the total input current is given by @xmath21 where @xmath22 index neurons .
the noise current @xmath23 has the same form as that for cortical neurons but with the noise intensity @xmath24 .
@xmath25 is a time - varying external input current which is injected to all sensory neurons . for each run of the simulation ,
the external input current is constructed by the following process .
let @xmath26 denote an ornstein - uhlenbeck process , which is described by @xmath27 where @xmath28 is a gaussian white noise with zero mean and unit variance , @xmath29 is a correlation time constant , and @xmath30 is a diffusion coefficient . the external input current @xmath25
is defined as @xmath31 parameter @xmath30 can be used to denote the intensity of the external input signal @xmath25 . in this work ,
we choose @xmath32 @xmath33 and @xmath34 ms . by its definition
, the external input current @xmath25 corresponds to a gaussian - distributed white noise low - pass filtered at 80 ms and half - wave rectified .
it should be noted that this type of external input current is widely used in the literature , in particular in the research papers which study the firing rate propagation ( van rossum et al .
2002 ; vogels and abbott 2005 ; wang and zhou 2009 ) . the synaptic interactions between neurons are implemented by using the modified conductance - based model .
our modeling methodology is inspired by the phenomenon of probabilistic transmitter release of the real biological synapses .
here we only introduce the model of unreliable excitatory synapses , because the propagation of neural activity is mainly examined in ure feedforward neuronal networks in this work .
the methods about how to model reliable excitatory synapses and unreliable inhibitory synapses will be briefly introduced in secs .
[ sec:3d ] and [ sec:3e ] , respectively .
the total synaptic current onto neuron @xmath6 is the linear sum of the currents from all incoming synapses , @xmath35 in this equation , the outer sum runs over all synapses onto this particular neuron , @xmath36 is the conductance from neuron @xmath37 to neuron @xmath6 , and @xmath38 mv is the reversal potential of the excitatory synapse . whenever the neuron @xmath37 emits a spike , an increment is assigned to the corresponding synaptic conductances according to the synaptic reliability parameter , which process is given by @xmath39 where @xmath40 denotes the synaptic reliability parameter of the synapse from neuron @xmath37 to neuron @xmath6 , and @xmath41 stands for the relative peak conductance of this particular excitatory synapse which is used to determine its strength .
for simplicity , we assume that @xmath42 , that is , the synaptic strength is identical for all excitatory connections .
parameter @xmath43 is defined as the successful transmission probability of spikes .
when a presynaptic neuron @xmath37 fires a spike , we let the corresponding synaptic reliability variables @xmath44 with probability @xmath43 and @xmath45 with probability @xmath46 . that is to say , whether the neurotransmitter is successfully released or not is in essence controlled by a bernoulli on - off process in the present work . in other time , the synaptic conductance decays by an exponential law : @xmath47 with a fixed synaptic time constant @xmath48 . in the case of synfire propagation , we choose @xmath49 ms , and in the case of firing rate propagation , we choose @xmath50 ms . we now introduce several useful measures used to quantitatively evaluate the performance of the two different propagation modes : the synfire mode and firing rate mode .
the propagation of synfire activity is measured by the survival rate and the standard deviation of the spiking times of the synfire packet ( gewaltig et al .
let us first introduce how to calculate the survival rate for the synfire propagation . in our simulations , we find that the synfire propagation can be divided into three types : the failed synfire propagation , the stable synfire propagation , as well as the synfire instability propagation ( for detail , see sec .
[ sec:3a ] ) . for neurons in each layer
, a threshold method is developed to detect the local highest `` energy '' region . to this end
, we use a 5 ms moving time window with 0.1 ms sliding step to count the number of spikes within each window . here a high energy region means that the number of spikes within the window is larger than a threshold @xmath51 . since we use a moving time window with small sliding step
, there might be a continuous series of windows contain more than 50 spikes around a group of synchronous spikes . in this work ,
we only select the first window which covers the largest number of spikes around a group of synchronous spikes as the local highest energy region .
we use the number of local highest energy region to determine which type of synfire propagation occurs .
if there is no local highest energy region detected in the final layer of the network , we consider it as the failed synfire propagation .
when two or more separated local highest energy regions are detected in one layer , we consider it as the synfire instability propagation .
otherwise , it means the occurrence of the stable synfire propagation . for each experimental setting
, we carry out the simulation many times .
the survival rate of the synfire propagation is defined as the ratio of the number of occurrence of the stable synfire propagation to the total number of simulations . in additional simulations
, it turns out that the threshold value @xmath52 can vary in a wide range without altering the results . under certain conditions
, noise can help the feedforward neuronal network produce the spontaneous spike packets , which promotes the occurrence of synfire instability propagation and therefore decreases the survival rate . for stable synfire propagation
, there exists only one highest energy region for neurons in each layer .
spikes within this region are considered as the candidate synfire packet , which might also contain a few spontaneous spikes caused by noise and other factors . in this work ,
an adaptive algorithm is introduced to eliminate spontaneous spikes from the candidate synfire packet .
suppose now that there is a candidate synfire packet in the @xmath4-th layer with the number of spikes it contains @xmath53 and the corresponding spiking times @xmath54 .
the average spiking time of the candidate synfire packet is therefore given by @xmath55 thus the standard deviation of the spiking times in the @xmath4-th layer can be calculated as follows : @xmath56 ^ 2}. \end{split } \label{eq:10}\ ] ] we remove the @xmath5-th spike from the candidate synfire packet if it satisfies : @xmath57 , where @xmath58 is a parameter of our algorithm .
we recompute the average spiking time as well as the standard deviation of the spiking times for the new candidate synfire packet , and repeat the above eliminating process , until no spike is removed from the new candidate synfire packet anymore .
we define the remaining spikes as the synfire packet , which is characterized by the final values of @xmath53 and @xmath59 .
parameter @xmath58 determines the performance of the proposed algorithm .
if @xmath58 is too large , the synfire packet will lose several useful spikes at its borders , and if @xmath58 is too small , the synfire packet will contain some noise data . in our simulations
, we found that @xmath60 can result in a good compromise between these two extremes .
it should be emphasized that our algorithm is inspired by the method given in ( gewaltig et al .
next , we introduce how to measure the performance of the firing rate propagation .
the performance of firing rate propagation is evaluated by combining it with a population code .
specifically , we compute how similar the population firing rates in different layers to the external input current @xmath25 ( van rossum et al .
2002 ; vogels and abbott 2005 ) . to do this ,
a 5 ms moving time window with 1 ms sliding step is also used to estimate the population firing rates @xmath61 for different layers as well as the smooth version of the external input current @xmath62 .
the correlation coefficient between the population firing rate of the @xmath4-th layer and external input current is calculated by @xmath63 \left[r_i(k)-\overline{r}_i\right]\right\rangle_t}{\sqrt{\left\langle \left[i_s(k+\tau)-\overline{i}_s\right]^2\right\rangle_t\left \langle\left[r_i(k)-\overline{r}_i\right]^2\right\rangle_t } } , \end{split } \label{eq:11}\ ] ] where @xmath64 denotes the average over time .
here we use the maximum cross - correlation coefficient @xmath65 to quantify the performance of the firing rate propagation in the @xmath4-th layer .
note that @xmath66 is a normalization measure and a larger value corresponds to a better performance . in all numerical simulations
, we use the standard euler - maruyama integration scheme to numerically calculate the aforementioned stochastic differential eqs .
( [ eq:1])-([eq:8 ] ) ( kloeden et al . 1994 ) .
the temporal resolution of integration is fixed at 0.02 ms for calculating the measures of the synfire mode and at 0.05 ms for calculating the measures of the firing rate mode , as the measurement of the synfire needs higher precise .
in additional simulations , we have found that further reducing the integration time step does not change our numerical results in a significant way .
for the synfire mode , all simulations are executed at least 100 ms to ensure that the synfire packet can be successfully propagated to the final layer of the considered network .
while studying the firing rate mode , we perform all simulations up to 5000 ms to collect enough spikes for statistical analysis .
it should be noted that , to obtain convincing results , we carry out several times of simulations ( at least 200 times for the synfire mode and 50 times for the firing rate mode ) for each experimental setting to compute the corresponding measures .
in this section , we report the main results obtained in the simulation .
we first systematically investigate the signal propagation in the ure feedforward neuronal networks .
then , we compare these results with those for the corresponding rre feedforward neuronal networks . finally , we further study the signal propagation in the urei feedforward neuronal networks . here
we study the role of unreliable synapses on the propagation of synfire packet in the ure feedforward neuronal networks . in the absence of noise
, we artificially let each sensory neuron fire and only fire an action potential at the same time ( @xmath67 and @xmath68 ms ) .
without loss of generality , we let all sensory neurons fire spikes at @xmath69 ms .
figure [ fig:2 ] shows four typical spike raster diagrams of propagating synfire activity .
note that the time scales in figs .
[ fig:2a]-[fig:2d ] are different .
the ure feedforward neuronal network with both small successful transmission probability and small excitatory synaptic strength badly supports the synfire propagation . in this case ,
due to high synaptic unreliability and weak excitatory synaptic interaction between neurons , the propagation of synfire packet can not reach the final layer of the whole network ( see fig . [ fig:2a ] ) .
for suitable values of @xmath43 and @xmath70 , we find that the synfire packet can be stably transmitted in the ure feedforward neuronal network .
moreover , it is obvious that the width of the synfire packet at any layer for @xmath71 is much narrower than that of the corresponding synfire packet for @xmath72 ( see figs .
[ fig:2b ] and [ fig:2c ] ) . at the same time , the transmission speed is also enhanced with the increasing of @xmath43 .
these results indicate that the neuronal response of the considered network is much more precise and faster for suitable large successful transmission probability .
however , our simulation results also reveal that a strong excitatory synaptic strength with large value of @xmath43 might destroy the propagation of synfire activity . as we see from fig .
[ fig:2d ] , the initial tight synfire packet splits into several different synfire packets during the transmission process . such phenomenon is called the `` synfire instability '' ( tetzlaff et al . 2002 ; tetzlaff et al .
2003 ) , which mainly results from the burst firings of several neurons caused by the strong excitatory synaptic interaction as well as the stochastic fluctuation of the synaptic connections .
+ + in fig .
[ fig:3a ] , we depict the survival rate of synfire propagation as a function of the successful transmission probability @xmath43 for different values of excitatory synaptic strength @xmath70 , with the noise intensity @xmath73 .
we find that each survival rate curve can be at least characterized by one corresponding critical probability @xmath74 . for small @xmath43 ,
due to low synaptic reliability , any synfire packet can not reach the final layer of the ure feedforward neuronal network .
once the successful transmission probability @xmath43 exceeds the critical probability @xmath74 , the survival rate rapidly transits from 0 to 1 , suggesting that the propagation of synfire activity becomes stable for a suitable high synaptic reliability . on the other hand , besides the critical probability @xmath74 , we find that the survival rate curve should be also characterized by another critical probability @xmath75 if the excitatory synaptic strength is sufficiently strong ( for example , @xmath76 ns in fig .
[ fig:3a ] ) . in this case ,
when @xmath77 , our simulation results show that the survival rate rapidly decays from 1 to 0 , indicating that the network fails to propagate the stable synfire packet again .
however , it should be noted that this does not mean that the synfire packet can not reach the final layer of the network in this situation , but because the excitatory synapses with both high reliability and strong strength lead to the occurrence of the redundant synfire instability in transmission layers . to systematically establish the limits for the appearance of stable synfire propagation as well as to check that
whether our previous results can be generalized within a certain range of parameters , we further calculate the survival rate of synfire propagation in the @xmath78 panel , which is shown in fig .
[ fig:3b ] .
as we see , the whole @xmath78 panel can be clearly divided into three regimes .
these regimes include the failed synfire propagation regime ( regime i ) , the stable synfire propagation regime ( regime ii ) , and the synfire instability propagation regime ( regime iii ) .
our simulation results reveal that transitions between these regimes are normally very fast and therefore can be described as a sharp transition .
the data shown in fig . [ fig:3b ] further demonstrate that synfire propagation is determined by the combination of both the successful transmission probability and excitatory synaptic strength . for a lower synaptic reliability , the ure
feedforward neuronal network might need a larger @xmath70 to support the stable propagation of synfire packet . in reality ,
not only the survival rate of the synfire propagation but also its performance is largely influenced by the successful transmission probability and the strength of the excitatory synapses . in figs .
[ fig:3c ] and [ fig:3d ] , we present the standard deviation of the spiking times of the output synfire packet @xmath79 for different values of @xmath43 and @xmath70 , respectively . note that here we only consider parameters @xmath43 and @xmath70 within the stable synfire propagation regime .
the results illustrated in fig .
[ fig:3c ] clearly demonstrate that the propagation of synfire packet shows a better performance for a suitable higher synaptic reliability .
for the ideal case @xmath80 , the ure feedforward neuronal network even has the capability to propagate the perfect synfire packet ( @xmath81 and @xmath82 ms ) in the absence of noise .
on the other hand , it is also found that for a fixed @xmath43 the performance of synfire propagation becomes better and better as the value of @xmath70 is increased ( see fig .
[ fig:3d ] ) .
the above results indicate that both high synaptic reliability and strong excitatory synaptic strength are able to help the ure feedforward neuronal network maintain the precision of neuronal response in the stable synfire propagation regime . up to now
, we only use the perfect initial spike packet ( @xmath67 and @xmath68 ms ) to evoke the synfire propagation .
this is a special case which is simplified for analysis , but it is not necessary to restrict this condition . to understand how a generalized spike packet is propagated through the ure feedforward neuronal network , we randomly choose @xmath83 neurons from the sensory layer , and let each of these neurons fire and only fire a spike at any moment according to a gaussian distribution with the standard deviation @xmath84 . in figs .
[ fig:4a ] and [ fig:4b ] , we plot the survival rate of the synfire propagation as a function of @xmath83 and @xmath84 for four different values of successful transmission probability , respectively . when the successful transmission probability is not too large ( for example , @xmath85 , 0.27 , and 0.3 in figs .
[ fig:4a ] and [ fig:4b ] ) , the synfire activity is well build up after several initial layers for sufficiently strong initial spike packet ( large @xmath83 and small @xmath84 ) , and then this activity can be successfully transmitted along the entire network with high survival rate . in this case ,
too weak initial spike packet ( small @xmath83 and large @xmath84 ) leads to the propagation of the neural activities becoming weaker and weaker with the increasing of layer number .
finally , the neural activities are stopped before they reach the final layer of the network . moreover , with the increasing of the successful transmission probability , neurons in the downstream layers will share more common synaptic currents from neurons in the corresponding upstream layers .
this means that neurons in the considered network have the tendency to fire more synchronously for suitable larger @xmath43 ( not too large ) . on the other hand , for sufficiently high synaptic reliability ( for instance , @xmath86 in figs .
[ fig:4a ] and [ fig:4b ] ) , a large @xmath83 or a suitable large @xmath84 may result in the occurrence of synfire instability , which also reduces the survival rate of the synfire propagation .
therefore , for a fixed @xmath70 , the ure feedforward neuronal network with suitable higher synaptic reliability has the ability to build up stable synfire propagation from a slightly weaker initial spike packet ( see figs .
[ fig:4a ] and [ fig:4b ] ) . figures [ fig:4c ] and [ fig:4d ] illustrate the values of @xmath53 and @xmath59 versus the layer number for different initial spike packets and several different intrinsic system parameters of the network ( the successful transmission probability @xmath43 and excitatory synaptic strength @xmath70 ) .
for each case shown in figs .
[ fig:4c ] and [ fig:4d ] , once the synfire propagation is successfully established , @xmath87 converges fast to the saturated value 100 and @xmath59 approaches to an asymptotic value .
although the initial spike packet indeed determines whether the synfire propagation can be established or not as well as influences the performance of synfire propagation in the first several layers , but it does not determine the value of @xmath59 in deep layers provided that the synfire propagation is successfully evoked .
for the same intrinsic system parameters , if we use different initial spike packets to evoke the synfire propagation , the value of @xmath59 in deep layers is almost the same for different initial spike packets ( see fig .
[ fig:4d ] ) .
the above results indicate that the performance of synfire propagation in deep layers of the ure feedforward neuronal network is quite stubborn , which is mainly determined by the intrinsic parameters of the network but not the parameters of the initial spike packet .
in fact , many studies have revealed that the synfire activity is governed by a stable attractor in the @xmath88 space ( diesmann et al .
1999 ; diesmann et al . 2001 ; diesmann 2002 ; gewaltig et al .
our above finding is a signature that the stable attractor of synfire propagation does also exist for the feedforward neuronal networks with unreliable synapses .
next , we study the dependence of synfire propagation on neuronal noise .
it is found that both the survival rate of synfire propagation and its performance are largely influenced by the noise intensity .
there is no significant qualitative difference between the corresponding survival rate curves in low synaptic reliable regime .
however , we find important differences between these curves for small @xmath70 in high synaptic reliable regime , as well as for large @xmath70 in intermediate synaptic reliable regime , that is , during the transition from the successful synfire regime to the synfire instability propagation regime ( see from fig .
[ fig:5a ] ) . for each case
, it is obvious that the top region of the survival rate becomes smaller with the increasing of noise intensity .
this is at least due to the following two reasons : ( i ) noise makes neurons desynchronize , thus leading to a more dispersed synfire packet in each layer . for relatively high synaptic reliability , a dispersed synfire packet has the tendency to increase the occurrence rate of the synfire instability .
( ii ) noise with large enough intensity results in several spontaneous neural firing activities at random moments , which also promote the occurrence of the synfire instability .
figure [ fig:5b ] presents the value of @xmath79 as a function of the noise intensity @xmath19 for different values of successful transmission probability @xmath43 .
as we see , the value of @xmath79 becomes larger and larger as the noise intensity is increased from 0 to 0.1 ( weak noise regime ) .
this is also due to the fact that the existence of noise makes neurons desynchronize in each layer .
however , although noise tends to reduce the synchrony of synfire packet , the variability of @xmath59 in deep layers is quite low ( data not shown ) .
the results suggest that , in weak noise regime , the synfire packet can be stably transmitted through the feedforward neuronal network with small fluctuation in deep layers , but displays slightly worse performance compared to the case of @xmath73 .
further increase of noise will cause many spontaneous neural firing activities which might significantly deteriorate the performance of synfire propagation .
however , it should be emphasized that , although the temporal spread of synfire packet tends to increase as the noise intensity grows , several studies have suggested that under certain conditions the basin of attraction of synfire activity reaches a maximum extent ( diesmann 2002 ; postma et al . 1996 ; boven and aertsen 1990 ) .
such positive effect of noise can be compared to a well known phenomenon called aperiodic stochastic resonance ( collins et al .
1995b ; collins et al .
1996 ; diesmann 2002 ) . in this subsection
, we examine the firing rate propagation in ure feedforward neuronal networks . to this end , we assume that all sensory neurons are injected to a same time - varying external current @xmath25 ( see sec . [ sec:2b ] for detail ) .
note now that the sensory neurons are modeled by using the integrate - and - fire neuron model in the study of the firing rate propagation .
the maximum cross - correlation coefficient between the smooth version of external input current @xmath62 and the population firing rate of sensory neurons @xmath89 for different noise intensities.,width=294 ] before we present the results of the firing rate propagation , let us first investigate how noise influences the encoding capability of sensory neurons by the population firing rate .
this is an important preliminary step , because how much input information represented by sensory neurons will directly influence the performance of firing rate propagation .
the corresponding results are plotted in figs .
[ fig:6 ] and [ fig:7 ] , respectively .
when the noise is too weak , the dynamics of sensory neurons is mainly controlled by the same external input current , which causes neurons to fire spikes almost at the same time ( see figs . [ fig:7a ] and [ fig:7b ] ) . in this case , the information of the external input current is poorly encoded by the population firing rate since the synchronous neurons have the tendency to redundantly encode the same aspect of the external input signal .
when the noise intensity falls within a special intermediate range ( about 0.5 - 10 ) , neuronal firing is driven by both the external input current and noise . with the help of noise ,
the firing rate is able to reflect the temporal structural information ( i.e. , temporal waveform ) of the external input current to a certain degree ( see figs .
[ fig:7c ] to [ fig:7f ] ) , and therefore @xmath90 has large value in this situation . for too large noise intensity , the external input current is almost drowned in noise , thus resulting that the input information can not be well read from the population firing rate of sensory neurons again . on the other hand
, sensory neurons can fire `` false '' spikes provided that they are driven by sufficiently strong noise ( as for example at @xmath91 ms in fig .
[ fig:7e ] ) .
although the encoding performance of the sensory neurons might be good enough in this case , our numerical simulations reveal that such false spikes will seriously reduce the performance of the firing rate propagation in deep layers , which will be discussed in detail in the later part of this section . by taking these factors into account
, we consider the noise intensity of sensory neurons to be within the range of 0.5 to 1 in the present work .
an example of the firing rate propagation in the ure feedforward network .
here we show the smooth version of the external input current @xmath62 , as and the population firing rates of layers 1 , 2 , 4 , 6 , 8 , and 10 , respectively .
system parameters are @xmath92 ns , @xmath93 , and @xmath94 . ]
figure [ fig:8 ] shows a typical example of the firing rate propagation . in view of the overall situation
, the firing rate can be propagated rapidly and basically linearly in the ure feedforward neuronal network .
however , it should be noted that , although the firing rates of neurons from the downstream layers tend to track those from the upstream layers , there are still several differences between the firing rates for neurons in two adjacent layers .
for example , it is obvious that some low firing rates may disappear or be slightly amplified in the first several layers , as well as some high firing rates are weakened to a certain degree during the whole transmission process .
therefore , as the neural activities are propagated across the network , the firing rate has the tendency to lose a part of local detailed neural information but can maintain a certain amount of global neural information . as a result , the maximum cross - correlation coefficient between @xmath62 and @xmath95 basically drops with the increasing of the layer number .
let us now assess the impacts of the unreliable synapses on the performance of firing rate propagation in the ure feedforward neuronal network .
figure [ fig:9a ] presents the value of @xmath96 versus the success transmission probability @xmath43 for various excitatory synaptic strengths . for a fixed value of @xmath70
, a bell - shaped @xmath96 curve is clearly seen by changing the value of successful transmission probability , indicating that the firing rate propagation shows the best performance at an optimal synaptic reliability level .
this is because , for each value of @xmath70 , a very small @xmath43 will result in the insufficient firing rate propagation due to low synaptic reliability , whereas a sufficiently large @xmath43 can lead to the excessive propagation of firing rate caused by burst firings .
based on above reasons , the firing rate can be well transmitted to the final layer of the ure feedforward neuronal network only for suitable intermediate successful transmission probabilities .
moreover , with the increasing of @xmath70 , the considered network needs a relatively small @xmath43 to support the optimal firing rate propagation . in fig .
[ fig:9b ] , we plot the value of @xmath96 as a function of the excitatory synaptic strength @xmath70 for different values of @xmath43 . here the similar results as those shown in fig .
[ fig:9a ] can be observed .
this is due to the fact that increasing @xmath70 and fixing the value of @xmath43 is equivalent to increasing @xmath43 and fixing the value of @xmath70 to a certain degree . according to the aforementioned results , we conclude that both the successful transmission probability and excitatory synaptic strength are critical for firing rate propagation in ure feedforward networks , and better choosing of these two unreliable synaptic parameters can help the cortical neurons encode neural information more accurately .
next , we examine the dependence of the firing rate propagation on neuronal noise .
the corresponding results are plotted in figs .
[ fig:10 ] and [ fig:11 ] , respectively .
figure [ fig:10a ] demonstrates that the noise of cortical neurons plays an important role in firing rate propagation .
noise of cortical neurons with appropriate intensity is able to enhance their encoding accuracy .
it is because appropriate intermediate noise , on the one hand , prohibits synchronous firings of cortical neurons in deep layers , and on the other hand , ensures that the useful neural information does not drown in noise .
however , the level of enhancement is largely influenced by the noise intensity of sensory neurons .
as we see , for a large value of @xmath24 , such enhancement is weakened to a great extent .
this is because slightly strong noise intensity of sensory neurons will cause these neurons to fire several false spikes and a part of these spikes can be propagated to the transmission layers .
if enough false spikes appear around the weak components of the external input current , these spikes will help the network abnormally amplify these weak components during the whole transmission process .
the aforementioned process can be seen clearly from an example shown in fig .
[ fig:11 ] . as a result
, the performance of the firing rate propagation might be seriously deteriorated in deep layers .
however , it should be noted that this kind of influence typically needs the accumulation of several layers .
our simulation results show that the performance of firing rate propagation can be well maintained or even becomes slightly better ( depending on the noise intensity of sensory neurons , see fig .
[ fig:6 ] ) in the first several layers for large @xmath24 ( see fig . [ fig:10b ] ) .
in fact , the above results are based on the assumption that each cortical neuron is driven by independent noise current with the same intensity .
our results can be generalized from the sensory layer to the transmission layers if we suppose that noise intensities for neurons in different transmission layers are different .
all these results imply that better tuning of the noise intensities of both the sensory and cortical neurons can enhance the performance of firing rate propagation in the ure feedforward neuronal network .
( color online ) an example of weak external input signal amplification .
system parameters are the successful transmission probability @xmath93 , excitatory synaptic strength @xmath92 ns , and noise intensities @xmath97 and @xmath98 , respectively.,width=453 ] from the numerical results depicted in secs .
[ sec:3a ] and [ sec:3b ] , we find that increasing @xmath70 with @xmath43 fixed has similar effects as increasing @xmath43 while keeping @xmath70 fixed for both the synfire mode and firing rate mode .
some persons might therefore postulate that the signal propagation dynamics in feedforward neuronal networks with unreliable synapses can be simply determined by the average amount of received neurotransmitter for each neuron in a time instant , which can be reflected by the product of @xmath99 . to check whether this is true ,
we calculate the measures of these two signal propagation modes as a function of @xmath100 for different successful transmission probabilities .
if this postulate is true , the ure feedforward neuronal network will show the same propagation performance for different values of @xmath43 at a fixed @xmath99 .
our results shown in figs . [ fig:12a]-[fig:12c ] clearly demonstrate that the signal propagation dynamics in the considered network can not be simply determined by the product @xmath99 or , equivalently , by the average amount of received neurotransmitter for each neuron in a time instant .
for both the synfire propagation and firing rate propagation , although the propagation performance exhibits the similar trend with the increasing of @xmath99 , the corresponding measure curves do not superpose in most parameter region for each case , and in some parameter region the differences are somewhat significant ( see figs .
[ fig:12b ] and [ fig:12c ] ) .
this is because of the stochastic effect of neurotransmitter release , that is , the unreliability of neurotransmitter release will add randomness to the system .
different successful transmission probabilities may introduce different levels of randomness , which will further affect the nonlinear spiking dynamics of neurons .
therefore , the ure feedforward neuronal network might display different propagation performance for different values of @xmath43 even at a fixed @xmath101 .
if we set the value of @xmath99 constant , a low synaptic reliability will introduce large fluctuations in the synaptic inputs . for small @xmath43
, according to the above reason , some neurons will fire spikes more than once in the large @xmath99 regime .
this mechanism increases the occurrence rate of the synfire instability .
thus , the ure feedforward neuronal network has the tendency to stop the stable synfire propagation for a small synaptic transmission probability ( see fig . [ fig:12a ] ) . on the other hand , a high synaptic reliability will introduce small fluctuations in the synaptic inputs for a fixed @xmath99 .
this makes neurons in the considered network fire spikes almost synchronously for a large @xmath43 , thus resulting the worse performance for the firing rate propagation in large @xmath99 regime ( see fig . [ fig:12c ] ) .
our above results suggest that the performance of the signal propagation in feedforward neuronal networks with unreliable synapses is not only purely determined by the change of synaptic parameters , but also largely influenced by the stochastic effect of neurotransmitter release . in this subsection
, we make comparisons on the propagation dynamics between the ure and the rre feedforward networks .
we first introduce how to generate a corresponding rre feedforward neuronal network for a given ure feedforward neuronal network .
suppose now that there is a ure feedforward neuronal network with successful transmission probability @xmath43 .
a corresponding rre feedforward neuronal network is constructed by using the connection density @xmath43 ( on the whole ) , that is , a synapse from one neuron in the upstream layer to one neuron in the corresponding downstream layer exists with probability @xmath43 . as in the ure feedforward neuronal
network given in sec .
[ sec:2a ] , there is no feedback connection from downstream neurons to upstream neurons and also no connection among neurons within the same layer in the rre feedforward neuronal network .
it is obvious that parameter @xmath43 has different meanings in these two different feedforward neuronal network models .
the synaptic interactions between neurons in the rre feedforward neuronal network are also implemented by using the conductance - based model ( see eqs .
( [ eq:6 ] ) and ( [ eq:7 ] ) for detail ) .
however , here we remove the constraint of the synaptic reliability parameter for the rre feedforward neuronal network , e.g. , @xmath44 in all cases .
a naturally arising question is what are the differences , if have , between the synfire propagation and firing propagation in ure feedforward neuronal networks and those in rre feedforward neuronal networks , although the numbers of active synaptic connections that taking part in transmitting spikes in a time instant are the same from the viewpoint of mathematical expectation .
( color online ) the difference between the synfire propagation in the ure feedforward neuronal network and the rre feedforward neuronal network . here
we show the value of survival rate as a function of @xmath84 for different network models . in all cases , @xmath73 and @xmath102 .
other system parameters are @xmath103 ns and @xmath104 ( dot : `` @xmath105 '' , and circle : `` @xmath106 '' ) , and @xmath107 ns and @xmath108 ( square : `` @xmath109 '' , and asterisk : `` @xmath110 '' ) .
each data point is calculated based on 500 different independent simulations with different random seeds.,width=294 ] for the synfire propagation , our simulation results indicate that , compared to the rre feedforward neuronal network , the ure feedforward neuronal network is able to suppress the occurrence of synfire instability to a certain degree , which can be seen clearly in fig .
[ fig:13 ] .
typically , this phenomenon can be observed in strong excitatory synaptic strength regime . due to the heterogeneity of connectivity
, some neurons in the rre feedforward neuronal network will have more input synaptic connections than the other neurons in the same network . for large value of @xmath70
, these neurons tend to fire spikes very rapidly after they received synaptic currents . if the width of the initial spike packet is large enough , these neurons might fire spikes again after their refractory periods , which are induced by a few spikes from the posterior part of the dispersed initial spike packet .
these spikes may increase the occurrence rate of the synfire instability . while in the case of ure feedforward neuronal network
, the averaging effect of unreliable synapses tends to prohibit neurons fire spikes too quickly .
therefore , under the equivalent parameter conditions , less neurons can fire two or more spikes in the ure feedforeard neuronal network . as a result ,
the survival rate of the synfire propagation for the ure feedforeard neuronal network is larger than that for the rre feedforward neuronal network ( see fig .
[ fig:13 ] ) , though not so significant . in further simulations ,
we find interesting results in small @xmath43 regime for the firing rate propagation . compared to the case of the ure feedforward neuronal network
, the rre feedforward neuronal network can better support the firing rate propagation in this small @xmath43 regime for strong excitatory synaptic strength ( see fig .
[ fig:14a ] ) .
it is because the long - time averaging effect of unreliable synapses at small @xmath43 tends to make neurons fire more synchronous spikes in the ure feedforward neuronal network through the homogenization process of synaptic currents .
however , with the increasing of @xmath43 , neurons in the downstream layers have the tendency to share more common synaptic currents from neurons in the corresponding upstream layers for both types of feedforward neuronal networks .
the aforementioned factor makes the difference of the performance of firing rate propagation between these two types of feedforward neuronal networks become small so that the @xmath96 curves almost coincide with each other for the case of @xmath86 ( see fig .
[ fig:14b ] ) .
although from the above results we can not conclude that unreliable synapses have advantages and play specific functional roles in signal propagation , not like those results shown in the previous studies ( goldman et al .
2002 ; goldman 2004 ) , at least it is shown that the signal propagation activities are different in ure and rre to certain degrees .
we should be cautioned when using random connections to replace unreliable synapses in modelling research .
however , it should be noted that the rre feedforward neuronal network considered here is just one type of diluted feedforward neuronal networks .
there exists several other possibilities to construct the corresponding diluted feedforward neuronal networks ( hehl et al . 2001 ) .
the similar treatments for these types of diluted feedforward neuronal networks require further investigation . in this subsection , we further study the signal propagation in the feedforward neuronal networks composed of both excitatory and inhibitory neurons connected in an all - to - all coupling fashion ( i.e. , the urei feedforward neuronal networks ) .
this study is necessary because real biological neuronal networks , especially mammalian neocortex , consist not only of excitatory neurons but also of inhibitory neurons .
the urei feedforward neuronal network studied in this subsection has the same topology as that shown in fig .
[ fig:1 ] . in simulations
, we randomly choose 80 neurons in each layer as excitatory and the rest of them as inhibitory , as the ratio of excitatory to inhibitory neurons is about @xmath111 in mammalian neocortex .
the dynamics of the unreliable inhibitory synapse is also modeled by using eqs .
( [ eq:6 ] ) and ( [ eq:7 ] ) .
the reversal potential of the inhibitory synapse is fixed at -75 mv , and its strength is set as @xmath112 , where @xmath113 is a scale factor used to control the relative strength of inhibitory and excitatory synapses . since the results of the signal propagation in urei feedforward neuronal networks are quite similar to those in ure feedforward neuronal networks , we omit most of them and only discuss the effects of inhibition in detail . figure [ fig:15 ] shows the survival rate of synfire propagation in the @xmath114 panel for three different excitatory synaptic strengths . depending on whether the synfire packet can be successfully and stably transmitted to the final layer of the urei feedforward neuronal network ,
the whole @xmath114 panel can also be divided into three regimes . for each considered case
, the network with both small successful transmission probability and strong relative strength of inhibitory and excitatory synapses ( failed synfire regime ) prohibits the stable propagation of the synfire activity . while in the case of high synaptic reliability and small @xmath113 ( synfire instability propagation regime )
, the synfire packet also can not be stably transmitted across the whole network due to the occurrence of synfire instability .
therefore , the urei feedforward neuronal network is able to propagate the synfire activity successfully in a stable way only for suitable combination of parameters @xmath43 and @xmath113 .
moreover , due to the competition between excitation and inhibition , the transitions between these different regimes can not be described as a sharp transition anymore , in particular , for large scale factor @xmath113 .
our results suggest that such non - sharp character is strengthen with the increasing of @xmath70 .
on the other hand , the partition of these different propagation regimes depends not only on parameters @xmath43 and @xmath113 but also on the excitatory synaptic strength @xmath70 . as the value of @xmath70
is decreased , both the synfire instability propagation regime and stable synfire propagation regime are shifted to the upper left of the @xmath114 panel at first , and then disappear one by one ( data not shown ) .
in contrast , a strong excitatory synaptic strength has the tendency to extend the areas of the synfire instability propagation regime , and meanwhile makes the stable synfire propagation regime move to the lower right of the @xmath114 panel .
( color online ) effect of inhibition on firing rate propagation .
here we show the value of @xmath96 as a function of scale factor @xmath113 for different excitatory synaptic strengths .
system parameters are @xmath93 , and @xmath115 in all cases . each data point is calculated based on 50 different independent simulations with different random seeds.,width=294 ] for the case of firing rate propagation , we plot the value of @xmath96 versus the scale factor @xmath113 for different excitatory synaptic strengths in fig .
[ fig:16 ] , with a fixed successful transmission probability @xmath93 .
when the excitatory synaptic strength is small ( for instance @xmath92 ns ) , due to weak excitatory synaptic interaction between neurons the urei feedforward neuronal network can not transmit the firing rate sufficiently even for @xmath116 . in this case ,
less and less neural information can be propagated to the final layer of the considered network with the increasing of @xmath113 .
therefore , @xmath96 monotonically decreases with the scale factor @xmath113 at first and finally approaches to a low steady state value .
note that here the low steady state value is purely induced by the spontaneous neural firing activities , which are caused by the additive gaussian white noise .
as the excitatory synaptic strength grows , more neural information can be successfully transmitted for small value of @xmath113 .
when @xmath70 is increased to a rather large value , such as @xmath117 ns , the coupling is so strong that too small scale factor will lead to the excessive propagation of firing rate .
however , in this case , the propagation of firing rate can still be suppressed provided that the relative strength of inhibitory and excitatory synapses is strong enough . as a result , there always exists an optimal scale factor to best support the firing rate propagation for each large excitatory synaptic strength ( see fig .
[ fig:16 ] ) .
if we fix the value of @xmath70 ( not too small ) , then the similar results can also be observed by changing the scale factor for a large successful transmission probability ( data not shown ) .
once again , this is due to the fact that increasing @xmath70 and fixing @xmath43 is equivalent to increasing @xmath43 and fixing @xmath70 to a certain degree .
the feedforward neuronal network provides us an effective way to examine the neural activity propagation through multiple brain regions .
although biological experiments suggested that the communication between neurons is more or less unreliable ( abeles 1991 ; raastad et al .
1992 ; smetters and zador 1996 ) , so far most relevant computational studies only considered that neurons transmit spikes based on reliable synaptic models .
in contrast to these previous work , we took a different approach in this work .
here we first built a 10-layer feedforward neuronal network by using purely excitatory neurons , which are connected with unreliable synapses in an all - to - all coupling fashion , that is , the so - called ure feedforward neuronal network in this paper .
the goal of this work was to explore the dependence of both the synfire propagation and firing rate propagation on unreliable synapses in the ure neuronal network , but was not limited this type of feedforward neuronal network .
our modelling methodology was motivated by experimental results showing the probabilistic transmitter release of biological synapses ( branco and staras 2009 ; katz 1966 ; katz 1969 ; trommershuser et al .
1999 ) . in the study of synfire mode
, it was observed that the synfire propagation can be divided into three types ( i.e. , the failed synfire propagation , the stable synfire propagation , and the synfire instability propagation ) depending on whether the synfire packet can be successfully and stably transmitted to the final layer of the considered network .
we found that the stable synfire propagation only occurs in the suitable region of system parameters ( such as the successful transmission probability and excitatory synaptic strength ) .
for system parameters within the stable synfire propagation regime , it was found that both high synaptic reliability and strong excitatory synaptic strength are able to support the synfire propagation in feedforward neuronal networks with better performance and faster transmission speed .
further simulation results indicated that the performance of synfire packet in deep layers is mainly influenced by the intrinsic parameters of the considered network but not the parameters of the initial spike packet , although the initial spike packet determines whether the synfire propagation can be evoked to a great degree .
in fact , this is a signature that the synfire activity is governed by a stable attractor , which is in agreement with the results given in ( diesmann et al .
1999 ; diesmann et al .
2001 ; diesmann 2002 ; gewaltig et al . 2001 ) . in the study of firing rate propagation
, our simulation results demonstrated that both the successful transmission probability and the excitatory synaptic strength are critical for firing rate propagation . too small successful transmission probability or too weak excitatory synaptic strength results in the insufficient firing rate propagation , whereas too large successful transmission probability or too strong excitatory synaptic strength has the tendency to lead to the excessive propagation of firing rate . theoretically speaking , better tuning of these two synaptic parameters
can help neurons encode the neural information more accurately .
on the other hand , neuronal noise is ubiquitous in the brain .
there are many examples confirmed that noise is able to induce many counterintuitive phenomena , such as stochastic resonance ( collins et al .
1995a ; collins et al .
1995b ; collins et al .
1996 ; chialvo et al . 1997 ; guo and li 2009 ) and coherence resonance ( pikovsky and kurths 1996 ; lindner and schimansky - geier 2002 ; guo and li 2009 ) .
here we also investigated how the noise influences the performance of signal propagation in ure feedforward neuronal networks .
the numerical simulations revealed that noise tends to reduce the performance of synfire propagation because it makes neurons desynchronized and causes some spontaneous neural firing activities .
further studies demonstrated that the survival rate of synfire propagation is also largely influenced by the noise .
in contrast to the synfire propagation , our simulation results showed that noise with appropriate intensity is able to enhance the performance of firing rate propagation in ure feedforward neuronal networks .
in essence , it is because suitable noise can help neurons in each layer maintain more accurate temporal structural information of the the external input signal .
note that the relevant mechanisms about noise have also been discussed in several previous work ( van rossum et al .
2002 ; masuda and aihara 2002 ; masuda and aihara 2003 ) , and our results are consistent with the findings given in these work .
furthermore , we have also investigated the stochastic effect of neurotransmitter release on the performance of signal propagation in the ure feedforward neuronal networks .
for both the synfire propagation and firing rate propagation , we found that the ure feedforward neuronal networks might display different propagation performance , even when their average amount of received neurotransmitter for each neuron in a time instant remains unchanged .
this is because the unreliability of neurotransmitter release will add randomness to the system .
different synaptic transmission probabilities will introduce different levels of stochastic effect , and thus might lead to different spiking dynamics and propagation performance .
these findings revealed that the signal propagation dynamics in feedforward neuronal networks with unreliable synapses is also largely influenced by the stochastic effect of neurotransmitter release .
finally , two supplemental work has been also performed in this paper . in the first work , we compared both the synfire propagation and firing rate propagation in ure feedforward neuronal networks with the results in corresponding feedforward neuronal networks composed of purely excitatory neurons but connected with reliable synapses in an random coupling fashion ( rre feedforward neuronal network ) .
our simulations showed that several different results exist for both the synfire propagation and firing rate propagation in these two different feedforward neuronal network models .
these results tell us that we should be cautioned when using random connections to replace unreliable synapses in modelling research . in the second work
, we extended our results to more generalized feedforward neuronal networks , which consist not only of the excitatory neurons but also of inhibitory neurons ( urei feedforward neuronal network ) .
the simulation results demonstrated that inhibition also plays an important role in both types of neural activity propagation , and better choosing of the relative strength of inhibitory and excitatory synapses can enhance the transmission capability of the considered network .
the results presented in this work might be more realistic than those obtained based on reliable synaptic models .
this is because the communication between biological neurons indeed displays the unreliable properties .
in real neural systems , neurons may make full use of the characteristics of unreliable synapses to transmit neural information in an adaptive way , that is , switching between different signal propagation modes freely as required .
further work on this topic includes at least the following two aspects : ( i ) since all our results are derived from numerical simulations , an analytic description of the synfire propagation and firing rate propagation in our considered feedforward neuronal networks requires investigation .
( ii ) intensive studies on signal propagation in the feedforward neuronal network with other types of connectivity , such as the mexican - hat - type connectivity ( hamaguchi et al . 2004 ; hamaguchi and aihara 2005 ) and the gaussian - type connectivity ( van rossum et al . 2002 ) , as well as in the feedforward neuronal network imbedded into a recurrent network ( aviel et al . 2003 ; vogels and abbott 2005 ; kumar et al . 2008 ) , from the unreliable synapses point of view are needed as well .
we thank feng chen , yuke li , qiuyuan miao , xin wei and qunxian zheng for valuable discussions on an early version of this manuscript .
we gratefully acknowledge the anonymous reviewers for providing useful comments and suggestion , which greatly improved our paper .
we also sincerely thank one reviewer for reminding us of a critical citation ( trommershuser and diesmann 2001 ) .
this work is supposed by the national natural science foundation of china ( grant no .
60871094 ) , the foundation for the author of national excellent doctoral dissertation of pr china , and the fundamental research funds for the central universities ( grant no . 1a5000 - 172210126 ) .
daqing guo would also like to thank the award of the ongoing best phd thesis support from the university of electronic science and technology of china .
boven , k. , h. , @xmath118 aertsen , a. , m. , h. , j. ( 1990 ) .
dynamics of activity in neuronal networks give rise to fast modulations of functional connectivity . in _
parallel processing and neural systems and computers _
( pp . 53 - 56 ) .
amsterdam : elsevier .
raastad , m. , storm , j. f. , @xmath118 andersen p. ( 1992 ) .
putative single quantum and single fibre excitatory postsynaptic currents show similar amplitude range and variability in rat hippocampal slices .
j. neurosci .
_ , 4 , 113 - 117 .
shinozaki , t. , okada , m. , reyes , a. d. , @xmath118 c@xmath0teau , h. ( 2010 ) .
flexible traffic control of the synfire - mode transmission by inhibitory modulation : nonlinear noise reduction .
e _ , 81 , 011913 .
trommershuser j. , marienhagen j. , @xmath118 zippelius a. ( 1999 ) , stochastic model of central synapses : slow diffusion of transmitter interacting with spatially distributed receptors and transporters .
_ j theor .
_ , 198 , 101 - 120 . | in this paper , we systematically investigate both the synfire propagation and firing rate propagation in feedforward neuronal network coupled in an all - to - all fashion . in contrast to most earlier work , where only reliable synaptic connections are considered , we mainly examine the effects of unreliable synapses on both types of neural activity propagation in this work .
we first study networks composed of purely excitatory neurons .
our results show that both the successful transmission probability and excitatory synaptic strength largely influence the propagation of these two types of neural activities , and better tuning of these synaptic parameters makes the considered network support stable signal propagation .
it is also found that noise has significant but different impacts on these two types of propagation .
the additive gaussian white noise has the tendency to reduce the precision of the synfire activity , whereas noise with appropriate intensity can enhance the performance of firing rate propagation .
further simulations indicate that the propagation dynamics of the considered neuronal network is not simply determined by the average amount of received neurotransmitter for each neuron in a time instant , but also largely influenced by the stochastic effect of neurotransmitter release .
second , we compare our results with those obtained in corresponding feedforward neuronal networks connected with reliable synapses but in a random coupling fashion . we confirm that some differences can be observed in these two different feedforward neuronal network models .
finally , we study the signal propagation in feedforward neuronal networks consisting of both excitatory and inhibitory neurons , and demonstrate that inhibition also plays an important role in signal propagation in the considered networks .
* keywords : * feedforward neuronal network , unreliable synapse , signal propagation , synfire chain , firing rate |
for each positive integer @xmath2 let @xmath1 denote the set of those integer partitions whose maximal hook lengths are strictly less than @xmath2 . recall that a partition @xmath5 covers
a partition @xmath6 in young s lattice means that the young diagram of @xmath6 is got by removing an inner corner box from the young diagram of @xmath5 .
consider the hasse diagram of the subposet of young s lattice on @xmath1 .
let us call @xmath7 the underlying undirected abstract graph .
( 2000,0)(-1000,800 ) ( 0,0 ) ( 0,-1000 ) ( 951,-309 ) ( 588,809 ) ( -588,809 ) ( -951,-309 ) ( 0,-724 ) ( 688,-224 ) ( 425,585 ) ( -425,585 ) ( -688,-224 ) ( 263,-362 ) ( 425,138 ) ( 0,447 ) ( -425,138 ) ( -263,-362 ) ( 0,-1000)(0,-724 ) ( 951,-309)(688,-224 ) ( 588,809)(425,585 ) ( -588,809)(-425,585 ) ( -951,-309)(-688,-224 ) ( 0,-724)(263,-362)(688,-224)(425,138)(425,585)(0,447)(-425,585 ) ( -425,138)(-688,-224)(-263,-362)(0,-724 ) ( 0,0)(263,-362 ) ( 0,0)(425,138 ) ( 0,0)(0,447 ) ( 0,0)(-425,138 ) ( 0,0)(-263,-362 ) this graph has a @xmath8-fold ( cyclic ) symmetry , and its full symmetry group is a dihedral group of order @xmath9 .
the phenomenon generalizes to all @xmath2 . for each positive integer @xmath2
the cyclic group of order @xmath2 acts faithfully on @xmath7 . mapping a partition to its dual (
also known as conjugate or transpose ) partition restricts to an involution @xmath10 , which is different from the identity if @xmath11 .
together with the cyclic symmetry of order @xmath2 it generates the full symmetry group @xmath12 , which is a dihedral group of order @xmath3 if @xmath11 .
the cyclic symmetries emerged as a byproduct of the work on abelian ideals in a borel subalgebra @xmath13 of a finite - dimensional complex simple lie algebra @xmath14 ( see @xcite ) , namely when @xmath14 has type @xmath15 .
here is a quick recapitulation .
let us use the convention that the nilradical of @xmath13 is the sum of the root spaces @xmath16 where @xmath17 runs over the _ positive _ roots .
then each abelian ideal @xmath18 is a sum of root spaces @xmath19 for a certain subset @xmath20 of the set of positive roots .
if @xmath21 denotes half the sum of the positive roots , then @xmath22 is the unique integral weight in the interior of an affine weyl group translate of the fundamental alcove ( for instance @xmath23 , where @xmath24 is the highest root , is such an integral weight in the interior of an alcove , which is adjacent to the fundamental alcove ) .
the union of all those alcoves that one gets if @xmath25 runs over all abelian ideals in @xmath13 , is exactly the fundamental alcove dilated by a factor of two . in the special case of @xmath26 the @xmath2-fold cyclic symmetry of the fundamental alcove , which is reflected in the fact that the affine dynkin diagram is an @xmath2-cycle , yields the cyclic symmetry acting on @xmath1 after identifying elements of @xmath1 in an evident way with abelian ideals and hence with those alcoves that lie in ( and tessellate ) the fundamental alcove dilated by a factor of two .
since integer partitions and young s lattice are so basic mathematical objects , it was natural to look for an independent approach to the cyclic symmetries .
this was realized in @xcite , and i reported on it in a talk `` a surprising result about young s lattice '' in the eth zurich algebra - topology seminar in october 2001 . in october 2007 ,
i mentioned the cyclic symmetries parenthetically in a talk `` some insights from cluster categories '' at indam in rome at a workshop organized by paolo papi and eric sommers . for recent developments
see @xcite .
on february 29 , 2012 , urs lang gave a talk `` injective hulls of metric spaces : old and new '' in the eth zurich geometry seminar .
in the introduction he showed the picture ( * ? ? ?
* top of p. 338 ) , which roughly looks as follows ( 6344,3509)(0,-10 ) ( 3022,2572)(4522,2722)(5122,1222 ) ( 3622,1072)(3022,2572 ) ( 3022,2572)(4522,2722)(5122,1222 ) ( 3622,1072)(3022,2572 ) ( 1822,2122)(3022,2572)(3622,1072 ) ( 2422,622)(1822,2122 ) ( 1822,2122)(3022,2572)(3622,1072 ) ( 2422,622)(1822,2122 ) ( 3322,2872)(4822,3022)(4522,2722 ) ( 3022,2572)(3322,2872 ) ( 3322,2872)(4822,3022)(4522,2722 ) ( 3022,2572)(3322,2872 ) ( 2722,3172)(3322,2872)(3022,2572 ) ( 2422,2872)(2722,3172 ) ( 2722,3172)(3322,2872)(3022,2572 ) ( 2422,2872)(2722,3172 ) ( 2422,2872)(3022,2572)(1822,2122 ) ( 1222,2422)(2422,2872 ) ( 2422,2872)(3022,2572)(1822,2122 ) ( 1222,2422)(2422,2872 ) ( 2422,2872)(2722,3172 ) ( 3022,2572)(3322,2872 ) ( 3622,1072)(2422,622 ) ( 3022,2572)(1822,2122 ) ( 1222,2422)(1822,2122 ) ( 1822,2122)(2422,622 ) ( 3022,2572)(4522,2722 ) ( 3322,2872)(4822,3022 ) ( 3622,1072)(5122,1222 ) ( 4522,2722)(4822,3022 ) ( 2722,3172)(3322,2872 ) ( 2722,3172)(2722,3472 ) ( 1222,2422)(22,3472 ) ( 4822,3022)(5572,3472 ) ( 2422,622)(2122,22 ) ( 5122,1222)(6322,622 ) ( 1222,2422)(2422,2872 ) ( 4522,2722)(5122,1222 ) ( 2422,2872)(3022,2572 ) ( 3022,2572)(3622,1072 ) and shows the injective hull ( of one of three possible types ) of a generic metric space with five points .
a drawing for the injective hull of a generic metric space with four points shows a rectangle with four `` antennas '' attached , one at each vertex ; for three points one gets a @xmath27-shaped tree ; for two points the injective hull is a line segment ; and the injective hull of a one - point space has one point .
hence for @xmath28 we recognize the geometric realizations of the graphs @xmath7 as the @xmath4-skeleta of the injective hulls of certain @xmath2-point metric spaces .
what about the @xmath29 situation ?
from @xcite the graph @xmath30 on @xmath31 vertices and with @xmath32 edges looks as in the following picture .
( 6396,7373)(0,-10 ) ( 6316,5479 ) ( 6316,1879 ) ( 3198,7279 ) ( 80,5479 ) ( 80,1879 ) ( 3198,79 ) ( 3198,6079 ) ( 5276,4879 ) ( 5276,2479 ) ( 3198,1279 ) ( 1120,4879 ) ( 1120,2479 ) ( 3198,4879 ) ( 4237,4279 ) ( 4237,3079 ) ( 3198,2479 ) ( 2159,3079 ) ( 2159,4279 ) ( 3198,3499 ) ( 3198,3859 ) ( 3718,5779 ) ( 3718,1579 ) ( 2678,5779 ) ( 2678,1579 ) ( 4756,5179 ) ( 4756,2179 ) ( 1640,5179 ) ( 1640,2179 ) ( 5276,4279 ) ( 5276,3079 ) ( 1120,4279 ) ( 1120,3079 ) ( 3198,4879)(2159,4279)(2159,3079 ) ( 3198,2479)(4237,3079)(4237,4279)(3198,4879 ) ( 3198,6079)(3198,7279 ) ( 5276,2479)(6316,1879 ) ( 3198,1279)(3198,79 ) ( 3198,6079)(1120,4879)(1120,2479 ) ( 3198,1279)(5276,2479)(5276,4879)(3198,6079 ) ( 1120,4879)(80,5479 ) ( 1120,2479)(80,1879 ) ( 5276,4879)(6316,5479 ) ( 3198,2479)(3198,3499 ) ( 2159,3079)(3198,3859)(4237,3079 ) ( 3198,3859)(3198,4879 ) ( 4237,4279)(5276,4279 ) ( 4237,3079)(5276,3079 ) ( 2159,4279)(1120,4279 ) ( 2159,3079)(1120,3079 ) ( 4237,3079)(4756,2179 ) ( 3198,2479)(2678,1579 ) ( 2159,3079)(1640,2179 ) ( 2159,4279)(3198,3499)(4237,4279 ) ( 3198,4879)(3718,5779 ) ( 3198,2479)(3718,1579 ) ( 3198,4879)(2678,5779 ) ( 4237,4279)(4756,5179 ) ( 2159,4279)(1640,5179 ) this graph is indeed isomorphic to the @xmath4-skeleton of the injective hull of the six - point metric space visualized in ( * ? ? ?
* fig . 1 on p. 570 ) or in ( * ? ? ?
* fig . 2 on p. 176 ) . in general ,
@xmath7 can be geometrically realized as the @xmath4-skeleton of the injective hull of an @xmath2-point metric space where the @xmath2 points form an orbit under an isometric cyclic action .
[ summarythm ] let @xmath33 .
consider a geometric realization of the graph @xmath7 as a metric space in which all edges have length @xmath4 and let @xmath34 be its boundary , that is , @xmath34 is the @xmath2-point subspace consisting of the @xmath2 pending vertices ( the tips of the `` antennas '' ) corresponding to the rectangular partitions @xmath35 ( for @xmath36 ) together with the empty partition .
then the @xmath4-skeleton of the injective hull @xmath37 is a geometric realization of the graph @xmath7 . for a proof see remark [ proofsummarythm ] .
let @xmath33 . we consider the discrete circle with @xmath2 sites @xmath38 .
two sites @xmath39 are neighbouring if @xmath40 or @xmath41 . expressed more uniformly , the two sites @xmath42 and @xmath43 are neighbouring for @xmath44 , where @xmath2 is identified with @xmath45 . the circle for @xmath46 with its @xmath47 sites .
( 12,0.7)(0,-0.3 ) ( 0,0)(1,0)5 ( 7,0)(1,0)5 ( 1,-0.1)(1,0)5(0,1)0.2 ( 8,-0.1)(1,0)5(0,1)0.2 ( 0,-0.1)(4,0)2(0,1)0.2 ( 7,-0.1)(4,0)2(0,1)0.2 ( 0.5,0.07)(0,0)[b]@xmath45 ( 1.5,0.07)(0,0)[b]@xmath4 ( 2.5,0.07)(0,0)[b]@xmath48 ( 3.5,0.07)(0,0)[b]@xmath49 ( 4.5,0.07)(0,0)[b]@xmath47 ( 6,0)(0,0)@xmath50 ( 7.5,0.07)(0,0)[b]@xmath45 ( 8.5,0.07)(0,0)[b]@xmath4 ( 9.5,0.07)(0,0)[b]@xmath48 ( 10.5,0.07)(0,0)[b]@xmath49 ( 11.5,0.07)(0,0)[b]@xmath45 next we consider a discrete mbius strip @xmath51 with @xmath52 embedded as its boundary .
namely , @xmath53 .
the mbius strip @xmath54 has @xmath55 sites represented by @xmath56 ( @xmath57 ) , and we may extend this to @xmath58 by identifying @xmath2 with @xmath45 .
the embedding @xmath59 is @xmath60 . by convention and slightly abusing the notation ,
we write @xmath61 for the _ class _ of the pair @xmath56 ( i.e. , its site in @xmath54 ) and hence @xmath62 . some definitions below ( for instance , in lemma [ lregular ] )
use the independence under the exchange @xmath63 for their well - definedness , something that we keep in mind with the tacit understanding . the mbius strip for @xmath46 with its @xmath9 sites .
( 6034,3649)(0,-10 ) ( 22,3012)(622,3612)(1222,3012 ) ( 1822,3612)(2422,3012)(3022,3612 ) ( 3622,3012)(4222,3612)(4822,3012 ) ( 2422,612)(22,3012 ) ( 2422,612)(3022,12)(5422,2412)(4822,3012 ) ( 1222,3012)(3622,612 ) ( 2422,3012)(4222,1212 ) ( 3622,3012)(4822,1812 ) ( 4822,3012)(5422,3612)(6022,3012)(5422,2412 ) ( 3622,3012)(1822,1212 ) ( 2422,3012)(1222,1812 ) ( 1222,3012)(622,2412 ) ( 622,3012)(0,0)@xmath64 ( 1822,3012)(0,0)@xmath65 ( 3022,3012)(0,0)@xmath66 ( 4222,3012)(0,0)@xmath67 ( 1222,2412)(0,0)@xmath68 ( 2422,2412)(0,0)@xmath69 ( 3622,2412)(0,0)@xmath70 ( 1822,1812)(0,0)@xmath71 ( 3022,1812)(0,0)@xmath72 ( 2422,1212)(0,0)@xmath73 ( 3022,612)(0,0)@xmath74 ( 3622,1212)(0,0)@xmath75 ( 4222,1812)(0,0)@xmath76 ( 4822,2412)(0,0)@xmath77 ( 5422,3012)(0,0)@xmath78 ( 6034,3649)(0,-10 ) ( 22,3012)(622,3612)(1222,3012 ) ( 1822,3612)(2422,3012)(3022,3612 ) ( 3622,3012)(4222,3612)(4822,3012 ) ( 2422,612)(22,3012 ) ( 2422,612)(3022,12)(5422,2412)(4822,3012 ) ( 1222,3012)(3622,612 ) ( 2422,3012)(4222,1212 ) ( 3622,3012)(4822,1812 ) ( 4822,3012)(5422,3612)(6022,3012)(5422,2412 ) ( 3622,3012)(1822,1212 ) ( 2422,3012)(1222,1812 ) ( 1222,3012)(622,2412 ) ( 622,3012)(0,0)@xmath64 ( 1822,3012)(0,0)@xmath65 ( 3022,3012)(0,0)@xmath66 ( 4222,3012)(0,0)@xmath67 ( 1222,2412)(0,0)@xmath68 ( 2422,2412)(0,0)@xmath69 ( 3622,2412)(0,0)@xmath70 ( 1822,1812)(0,0)@xmath71 ( 3022,1812)(0,0)@xmath72 ( 2422,1212)(0,0)@xmath73 ( 3022,612)(0,0)@xmath64 ( 3622,1212)(0,0)@xmath68 ( 4222,1812)(0,0)@xmath71 ( 4822,2412)(0,0)@xmath73 ( 5422,3012)(0,0)@xmath64 we need to familiarize ourselves with some notations for the geography of the mbius strip @xmath54 .
let us first have a local inspection . *
each pair of adjacent sites in the mbius strip is represented ( in at least one way ) as + ( 7,3 ) ( 0,2)(1,-1)3(1,1)1 ( 0,2)(1,1)2(1,-1)2 ( 1,2)(0,0)@xmath56 ( 2,1)(0,0)@xmath79 ( 3,1.5)(0,0)[l]@xmath80or + ( 6,3 ) ( 0,1)(1,-1)2(1,1)2 ( 0,1)(1,1)3(1,-1)1 ( 1,1)(0,0)@xmath56 ( 2,2)(0,0)@xmath81 ( 3,1.5)(0,0)[l]@xmath82 + the pairs on the right for @xmath83 are already represented by the pairs of sites @xmath84 and @xmath85 depicted on the left . *
each triple of sites consisting of a site adjacent to two sites at the boundary of the mbius strip can be represented as + ( 8,3 ) ( 0,2)(1,1)1 ( 1,1)(1,-1)2(1,1)2 ( 0,2)(1,1)2(1,-1)2 ( 3,3)(1,-1)1 ( 1,2)(0,0)@xmath86 ( 2,1)(0,0)@xmath87 ( 3,2)(0,0)@xmath88 ( 4,1)(0,0)[l]@xmath89 * each @xmath48@xmath90@xmath48 square of sites in the mbius strip can be represented as + ( 8,4 ) ( 0,2)(1,-1)3(1,1)2 ( 0,2)(1,1)3(1,-1)2 ( 1,2)(0,0)@xmath56 ( 2,1)(0,0)@xmath79 ( 2,3)(0,0)@xmath81 ( 3,2)(0,0)@xmath91 ( 4,1)(0,0)[l]@xmath92 + ( for @xmath93 this square has only three different sites . ) as regards global objects in the geography of @xmath54 , it will soon become evident that homotopically nontrivial loops are important to consider .
let @xmath94 adjacent to @xmath95 adjacent to @xmath96 adjacent to @xmath97 adjacent to @xmath98 adjacent to @xmath99 be such a loop consisting of @xmath2 sites ( see ( [ exampleloop ] ) for an example ) .
[ shortnontrivialloops ] let @xmath100 note that for @xmath101 with @xmath102 the sites in @xmath103 can not realize a homotopically nontrivial loop in the mbius strip .
[ exhaustx ] since for @xmath104 the sites in @xmath103 realize a homotopically nontrivial loop , the following property holds : for each @xmath104 @xmath105 lemma [ lregular ] below gives more precise information .
[ loopsandpartitions ] there is a bijection @xmath106 in particular , @xmath107 .
the loop @xmath108 will be referred to as the ( loop associated with the ) _ outer rim _ of @xmath5 .
represent the sites of @xmath54 in a triangular shape with sites @xmath56 ( @xmath109 ) .
the three corners @xmath64 `` upper left '' , @xmath110 `` bottom '' , and @xmath111 `` upper right '' of this triangular shape all represent the same site @xmath112 , and @xmath113 . for @xmath104
we obtain the corresponding partition @xmath5 such that the sites below the sites contained in @xmath103 ( in the triangular shape ) are just the boxes of the young diagram ( drawn in russian convention ) of @xmath5 .
so @xmath103 can be considered as the outer rim of @xmath5 .
( if @xmath5 has maximal hook length @xmath114 , we could consider @xmath5 as having its nonzero parts and in addition @xmath115 zero parts . ) here is an example with @xmath116 for the partition @xmath117 .
( 12034,6659)(0,-10 ) ( 3622,2422)(4822,3622)(5422,3022 ) ( 6022,3622)(7222,2422)(8422,3622 ) ( 9022,3022)(6022,22)(3622,2422 ) ( 3622,2422)(4822,3622)(5422,3022 ) ( 6022,3622)(7222,2422)(8422,3622 ) ( 9022,3022)(6022,22)(3622,2422 ) ( 3022,3022)(4822,4822)(5422,4222 ) ( 6022,4822)(7222,3622)(8422,4822 ) ( 9622,3622)(9022,3022)(8422,3622 ) ( 7222,2422)(6022,3622)(5422,3022 ) ( 4822,3622)(3622,2422)(3022,3022 ) ( 3022,3022)(4822,4822)(5422,4222 ) ( 6022,4822)(7222,3622)(8422,4822 ) ( 9622,3622)(9022,3022)(8422,3622 ) ( 7222,2422)(6022,3622)(5422,3022 ) ( 4822,3622)(3622,2422)(3022,3022 ) ( 3622,3022)(0,0)@xmath118 ( 4222,3622)(0,0)@xmath119 ( 4822,4222)(0,0)@xmath120 ( 5422,3622)(0,0)@xmath121 ( 6022,4222)(0,0)@xmath122 ( 6622,3622)(0,0)@xmath123 ( 7222,3022)(0,0)@xmath124 ( 7822,3622)(0,0)@xmath125 ( 8422,4222)(0,0)@xmath126 ( 9022,3622)(0,0)@xmath127 ( 4822,1222)(9622,6022 ) ( 4222,1822)(8422,6022 ) ( 3022,3022)(6022,6022 ) ( 2422,3622)(4822,6022 ) ( 1822,4222)(3622,6022 ) ( 1222,4822)(2422,6022 ) ( 622,5422)(1222,6022 ) ( 1222,6022)(6622,622 ) ( 2422,6022)(7222,1222 ) ( 3622,6022)(7822,1822 ) ( 4822,6022)(8422,2422 ) ( 6022,6022)(9022,3022 ) ( 8422,6022)(10222,4222 ) ( 9622,6022)(10822,4822 ) ( 10822,6022)(11422,5422 ) ( 5422,622)(6022,22)(12022,6022 ) ( 11422,6622)(10822,6022 ) ( 22,6022)(622,6622)(1222,6022 ) ( 1822,6622)(2422,6022)(3022,6622 ) ( 3622,6022)(4222,6622)(4822,6022 ) ( 5422,6622)(6022,6022)(6622,6622 ) ( 7222,6022)(7822,6622)(8422,6022 ) ( 9022,6622)(9622,6022)(10222,6622 ) ( 10822,6022)(5422,622)(22,6022 ) ( 7222,6022)(9622,3622 ) ( 3622,2422)(7222,6022 ) the outer rim of @xmath5 is @xmath128 here is an encoding of @xmath129 in terms of a graph with vertex set @xmath130 : there is an edge between @xmath42 and @xmath131 if and only if @xmath132 . ( 7120,7270)(0,-10 ) ( 6698,3560)(5926,1439)(3971,310 ) ( 5926,1439)(297,2431)(1748,702 ) ( 1748,6418 ) ( 1748,6418)(297,4689 ) ( 1748,6418)(3971,6810)(5926,5681 ) ( 3971,6810)(5926,1439 ) ( 6998,3560)(0,0)@xmath45 ( 6156,5874)(0,0)@xmath47 ( 4023,7105)(0,0)@xmath133 ( 1598,6678)(0,0)@xmath49 ( 15,4791)(0,0)@xmath134 ( 15,2329)(0,0)@xmath48 ( 1598,442)(0,0)@xmath135 ( 4023,15)(0,0)@xmath4 ( 6156,1246)(0,0)@xmath8 note that in this way we get a connected graph with @xmath2 vertices and @xmath2 edges , hence with exactly one cycle ( which can be a graph - theoretic loop ) . from this graph
we can easily read off a bijection @xmath136 such that if @xmath137 , then @xmath138 , so that remark [ exhaustx ] follows as a corollary . in the example above there is no choice for @xmath139 , @xmath140 , @xmath141 , @xmath142 ; for the vertices that belong to the cycle @xmath143 , there are exactly two choices : @xmath144 for general @xmath108 there are always two choices for the bijection @xmath21 except if the associated graph has a graph - theoretic loop , in which case there is a unique such bijection @xmath21 .
note also that the cycle always has odd length .
in fact , the edges of the cycle correspond to the sites where the loop @xmath108 turns , and the number of those sites is the number of outer corners ( if the maximal hook length of @xmath5 is strictly less than @xmath145 ) or the number of outer corners minus @xmath48 ( if the maximal hook length of @xmath5 is @xmath145 ) plus the number of inner corners of @xmath5 .
the result then follows from @xmath146 in the example the outer corners of @xmath5 are at @xmath121 and @xmath124 as well as at @xmath118 and @xmath127 ( as a site in @xmath54 we have @xmath147 , and @xmath108 does not turn there ) ; and the inner corners of @xmath5 are at @xmath148 [ @xmath149 , @xmath150 [ @xmath151 , and @xmath152 [ @xmath153 , where [ @xmath154 marks the corresponding site where @xmath108 turns around the inner corner @xmath79 of @xmath5 .
let us complete the proof of proposition [ loopsandpartitions ] . in general , to enumerate all elements @xmath103 of @xmath155 , we count separately for @xmath156 those @xmath103 that contain both @xmath157 and @xmath158
. there are @xmath159 chains of length @xmath2 consisting of successively adjacent sites starting at @xmath158 and ending at @xmath160 ( all inside the triangular shape ) .
a summation over @xmath131 gives a total of @xmath161 possibilities .
[ lregular ] for each site @xmath162 let @xmath163 with @xmath164 . or in other words , the vector @xmath165 has entries @xmath4 at positions @xmath42 and @xmath131 if @xmath166 or has an entry @xmath48 at position @xmath42 if @xmath167 , and all other entries @xmath45 . for each @xmath168 consider the @xmath2@xmath90@xmath2 matrix @xmath169 then @xmath170 is regular , in fact , its determinant is @xmath171
we start with @xmath172 .
the matrix @xmath173 is triangular with its @xmath64 diagonal entry @xmath48 and the other diagonal entries @xmath4 , hence with determinant @xmath48 .
we proceed by induction . if @xmath174 , then @xmath5 covers a partition @xmath6 with outer rim @xmath175 .
the situation can be depicted as follows : @xmath176 @xmath177 @xmath178 + where @xmath179 , @xmath180 , @xmath181 , and @xmath182 .
the young diagram of @xmath5 is got by adding the box at @xmath183 to the young diagram of @xmath6 .
we can write @xmath184 . since @xmath185 , the determinants of the matrices @xmath170 and @xmath186 are equal up to a sign .
( using the fact that there are exactly one or two distinguished bijections @xmath136 ( with @xmath138 if @xmath137 ) and that if there are two such bijections , then they differ only by an odd - length cyclic ( hence even ) permutation , we could remove the ambiguity in the sign of the determinant . )
[ cyclicaction ] the cyclic action that is given by translation in the mbius strip @xmath187 , @xmath188 induces an action on @xmath155 and hence on the set of partitions with maximal hook length @xmath189 .
this corresponds to the diagonal sliding operation described in @xcite and will be recalled later ( see theorem [ zerocells ] ) .
a metric space @xmath190 is called _ injective _ if every @xmath4-lipschitz ( i.e. , distance non - increasing ) map @xmath191 from a subspace @xmath183 of any metric space @xmath52 can be extended to a @xmath4-lipschitz map @xmath192 .
it was first proved by isbell @xcite that for every metric space @xmath52 there is an injective metric space @xmath193 such that @xmath52 embeds isometrically into @xmath193 ; and given such an embedding @xmath194 , there is a unique smallest injective subspace @xmath195 containing @xmath196 .
let @xmath197 where @xmath198 extends @xmath199 as a @xmath4-lipschitz map .
this space @xmath200 ( or more properly the isometric embedding @xmath201 ) is called an _ injective hull _ of @xmath52 .
if @xmath202 is another injective hull of @xmath52 , then there is an isometry @xmath203 with @xmath204 .
the injective hull @xmath200 of a finite metric space @xmath52 with metric @xmath205 can be realized as the polyhedral complex that consists of the bounded faces of the polyhedron @xmath206{dr2 } ) } \label{modelinjhull } { \operatorname{e}}(x)&:=\bigl\{f\in\delta(x)\bigm|\forall x\in x\,\exists y\in x : f(x)+f(y)=d(x , y)\bigr\}.\end{aligned}\ ] ] the distance between two functions @xmath207 is @xmath208 , and @xmath52 embeds into @xmath200 by @xmath209 .
we consider the metric space @xmath38 with metric @xmath210 ( for @xmath44 ; the formula holds also if we insert @xmath211 and identify @xmath2 with @xmath45 ) and extended cyclically , that is , @xmath212 .
note that if we let @xmath45 correspond to the empty partition and @xmath213 to the rectangular partition @xmath35 , then this is the metric induced from the hasse graph ( all edges of length @xmath4 ) of young s lattice .
in fact , starting with the young diagram of the partition @xmath35 , we successively remove boxes till we get @xmath214 ( for @xmath215 ) and then successively add boxes till we obtain @xmath216 .
( 5124,3942)(0,-10 ) ( 1018.066,551.360)(912.000,615.000)(975.640,508.934 ) ( 912,615)(1512,15 ) ( 1405.934,78.640)(1512.000,15.000)(1448.360,121.066 ) ( 2775.640,121.066)(2712.000,15.000)(2818.066,78.640 ) ( 2712,15)(4212,1515 ) ( 4148.360,1408.934)(4212.000,1515.000)(4105.934,1451.360 ) ( 312,2415)(1812,3915)(3612,2115 ) ( 2112,615)(312,2415 ) ( 1512,1215)(4212,3915)(4812,3315 ) ( 2112,615)(1512,1215 ) ( 118.066,2051.360)(12.000,2115.000)(75.640,2008.934 ) ( 12,2115)(1812,315 ) ( 1705.934,378.640)(1812.000,315.000)(1748.360,421.066 ) ( 2475.640,421.066)(2412.000,315.000)(2518.066,378.640 ) ( 2412,315)(5112,3015 ) ( 5048.360,2908.934)(5112.000,3015.000)(5005.934,2951.360 ) ( 3612,515)(0,0)[lb]@xmath42 ( 712,75)(0,0)@xmath217 ( 312,1215)(0,0)[t]@xmath218 ( 4612,2115)(0,0)[lb]@xmath131 the total number of required moves is @xmath219 [ embedfx ] we embed @xmath220 by @xmath221 this is of course well - defined and @xmath222 . note also that the extension commutes with affine combinations , that is , for @xmath223 and @xmath224 with @xmath225 we have @xmath226 this applies in particular to convex combinations , where all the coefficients @xmath227 are nonnegative and sum up to @xmath4 .
let us rewrite the definitions ( [ polyhedron ] ) and ( [ modelinjhull ] ) as in the following definition .
[ injhullx ] @xmath228 [ fliftrecursion ] let @xmath229 and let @xmath230 be its extension to @xmath54 as in definition [ embedfx ]
. for any @xmath48@xmath90@xmath48 square of sites @xmath231 we have + @xmath232 recall from section [ geography ] how to represent the sites in such triples and quadruples . for the triples
let @xmath233 , @xmath234 , @xmath235 and compute @xmath236[l]{$2\tilde f(a)+n-\tilde f(b)-\tilde f(c)$}\\ & = f(j)+f(j+1)-(n-1)+n - f(j)-f(j+1)=1 \intertext{and for the quadruples let $ a=(j , k+1)$ , $ b=(j , k)$ , $ c=(j+1,k+1)$ , $ d=(j+1,k)$ and compute } \makebox[1cm][l]{$\tilde f(a)+\tilde f(d)-\tilde f(b)-\tilde f(c)$}\\ & = \tfrac12\bigl(f(j)+f(k+1)-(k - j+1)(n - k+j-1)\bigr)\\ & \quad\,{}+\tfrac12\bigl(f(j+1)+f(k)-(k - j-1)(n - k+j+1)\bigr)\\ & \quad\,{}-\tfrac12\bigl(f(j)+f(k)-(k - j)(n - k+j)\bigr)\\ & \quad\,{}-\tfrac12\bigl(f(j+1)+f(k+1)-(k - j)(n - k+j)\bigr)=1.\\[-12mm]\end{aligned}\ ] ] [ hldjk ] for a site @xmath162 consider the affine hyperplane @xmath237 with @xmath238 the extension of @xmath239 to @xmath54 as in definition [ embedfx ] . for a set of sites @xmath101 consider the intersection of the affine hyperplanes @xmath240 the next proposition will be superseded by theorem [ zerocells ] .
let @xmath241 and let @xmath242 be the outer rim ( see proposition [ loopsandpartitions ] ) of the empty partition @xmath243 .
then * @xmath244 * for @xmath245 , say @xmath246 with @xmath247 ( or even @xmath109 ) , we have @xmath248 * @xmath249 we have @xmath250 and hence for @xmath247 @xmath251 since @xmath252 this implies in particular @xmath253 and by the regularity expressed in lemma [ lregular ] ( here we need only the base case in its inductive proof ) we conclude that @xmath244 .
the ( in)equalities ( [ fvaluesempty ] ) show that @xmath254 , and to conclude that @xmath249 just take @xmath255 in ( [ modelinjhull ] ) .
[ zerocells ] consider the @xmath2-point metric space @xmath256 with metric @xmath212 .
then the @xmath45-faces of its injective hull @xmath200 , realized as in definition [ injhullx ] , are precisely @xmath257 where @xmath5 runs through the partitions in @xmath1 , and @xmath108 is the outer rim of @xmath5 ( as defined in proposition [ loopsandpartitions ] ) .
the solutions @xmath258 have more explicit descriptions : 1 .
@xmath259 is determined by the following recursive procedure : @xmath250 ; and if @xmath5 covers @xmath6 such that @xmath260 for some pair @xmath56 with @xmath261 , then @xmath262 2 .
@xmath259 can be described directly via @xmath263 and by using the cyclic action .
for @xmath264 let @xmath265 ( for the empty partition put @xmath266 , @xmath267 ) .
then @xmath268 is the cyclic action alluded to in remark [ cyclicaction ] .
its inverse @xmath269 ( note the intended coincidence with the common notation for an auslander - reiten translation ) is then @xmath270 and explicitly @xmath271 ( for the empty partition put @xmath272 , @xmath273 ) .
the formula for @xmath259 is @xmath274 let @xmath275 . by lemma [ lregular ] the @xmath2 affine hyperplanes @xmath276 ( definition [ hldjk ] ) intersect in a single point @xmath259 .
for a site @xmath277 consider the partition @xmath278 which is defined as follows : its young diagram has the shape that is bordered by @xmath108 and whose `` bottom box '' is at the position of the site @xmath279 .
for @xmath280 let @xmath278 be the empty partition .
( 3774,2139)(0,-10 ) ( 1512,162)(1662,312)(1812,162 ) ( 12,1362)(462,1812)(612,1662 ) ( 912,1962)(1212,1662)(1362,1812 ) ( 1812,1362)(2112,1662)(2262,1512 ) ( 2712,1962)(3012,1662)(3462,2112 ) ( 3762,1812)(3612,1662)(3462,1812 ) ( 3012,1362)(2712,1662)(2262,1212 ) ( 2112,1362)(1812,1062)(1362,1512 ) ( 1212,1362)(912,1662)(612,1362 ) ( 462,1512)(162,1212)(12,1362 ) ( 12,1362)(462,1812)(612,1662 ) ( 912,1962)(1212,1662)(1362,1812 ) ( 1812,1362)(2112,1662)(2262,1512 ) ( 2712,1962)(3012,1662)(3462,2112 ) ( 3762,1812)(3612,1662)(3462,1812 ) ( 3012,1362)(2712,1662)(2262,1212 ) ( 2112,1362)(1812,1062)(1362,1512 ) ( 1212,1362)(912,1662)(612,1362 ) ( 462,1512)(162,1212)(12,1362 ) ( 312,1362)(1662,12)(3012,1362 ) ( 1662,162)(0,0)@xmath279 ( 1662,800)(0,0)shape @xmath278 ( -75,1125)(0,0)@xmath108 ( 3774,2139)(0,-10 ) ( 1512,162)(1662,312)(1812,162 ) ( 12,1362)(462,1812)(612,1662 ) ( 912,1962)(1212,1662)(1362,1812 ) ( 1812,1362)(2112,1662)(2262,1512 ) ( 2712,1962)(3012,1662)(3462,2112 ) ( 3762,1812)(3612,1662)(3462,1812 ) ( 3012,1362)(2712,1662)(2262,1212 ) ( 2112,1362)(1812,1062)(1362,1512 ) ( 1212,1362)(912,1662)(612,1362 ) ( 462,1512)(162,1212)(12,1362 ) ( 12,1362)(462,1812)(612,1662 ) ( 912,1962)(1212,1662)(1362,1812 ) ( 1812,1362)(2112,1662)(2262,1512 ) ( 2712,1962)(3012,1662)(3462,2112 ) ( 3762,1812)(3612,1662)(3462,1812 ) ( 3012,1362)(2712,1662)(2262,1212 ) ( 2112,1362)(1812,1062)(1362,1512 ) ( 1212,1362)(912,1662)(612,1362 ) ( 462,1512)(162,1212)(12,1362 ) ( 312,1362)(1662,12)(3012,1362 ) ( 612,1362)(1662,312)(2862,1512 ) ( 1362,312)(1662,612)(1962,312 ) ( 1662,162)(0,0)@xmath183 ( 1512,312)(0,0)@xmath281 ( 1812,312)(0,0)@xmath282 ( 1662,462)(0,0)@xmath283 ( -75,1125)(0,0)@xmath108 in particular , we recover @xmath5 as @xmath284 for a quadruple @xmath285 of sites as depicted on the right side in the illustration above we get of course by considering the areas @xmath286 in fact , for @xmath280 both sides in ( [ numberofsites ] ) vanish .
the formula ( [ abcdformula ] ) computes the values @xmath287 for all @xmath288 recursively from the values @xmath289 for @xmath280 . on the other hand , by lemma [ fliftrecursion ]
, @xmath290 satisfies the same recursive formula .
hence ( [ numberofsites ] ) holds true . from ( [ recoverlambda ] ) and ( [ numberofsites ] ) we get the formula ( [ fdirectly ] ) for the case @xmath291 , namely ,
@xmath263 . more generally , if @xmath292 with dual partition @xmath293 , then @xmath294 the formula ( [ fdirectly ] ) follows from @xmath295 , where @xmath296 is the translation @xmath188 in the mbius strip , together with ( [ recoverlambda ] ) and ( [ numberofsites ] ) .
before we continue with the first part of theorem [ zerocells ] , we use the geometric interpretation to compare @xmath297 and @xmath298 if @xmath5 covers @xmath6 . their outer rims @xmath108 and @xmath175 satisfy @xmath299 the following picture illustrates the situation .
( 4824,5139)(0,-10 ) ( 1812,2712)(2112,3012)(2412,2712 ) ( 2112,2412)(1812,2712 ) ( 1812,2712)(2112,3012)(2412,2712 ) ( 2112,2412)(1812,2712 ) ( 2412,2712)(2712,3012)(3012,2712 ) ( 2712,2412)(2412,2712 ) ( 2412,2712)(2712,3012)(3012,2712 ) ( 2712,2412)(2412,2712 ) ( 312,4812)(4812,312)(4512,12 ) ( 3912,612)(3612,312)(3912,12)(4512,612 ) ( 312,4812)(612,5112)(1212,4512 ) ( 1512,4812)(1212,5112)(612,4512 ) ( 1212,312)(1512,12)(1812,312 ) ( 2112,12)(2412,312)(2712,12 ) ( 3012,312)(3312,12)(3612,312 ) ( 1512,4812)(1812,5112)(2112,4812 ) ( 2412,5112)(2712,4812)(3012,5112)(3312,4812 ) ( 312,612)(912,12)(1212,312 ) ( 912,612)(312,12)(12,312 ) ( 4512,4812)(4212,5112)(3612,4512 ) ( 3312,4812)(3612,5112)(4212,4512 ) ( 2112,2412)(2412,2112)(2712,2412 ) ( 2112,3012)(2412,3312)(2712,3012 ) ( 612,4812)(0,0)@xmath300 ( 4212,4812)(0,0)@xmath301 ( 312,312)(0,0)@xmath302 ( 4512,312)(0,0)@xmath86 ( 2412,2412)(0,0)@xmath303 ( 2412,3012)(0,0)@xmath304 ( 3000,3012)(0,0)[l]@xmath305 ( 3000,2412)(0,0)[l]@xmath306 ( 2412,4012)(0,0)@xmath307 ( 2412,3712)(0,0)for @xmath279 in this sector ( 2412,1412)(0,0)@xmath308 ( 2412,1112)(0,0)for @xmath279 in this sector ( -50,3012)(0,0)[l]@xmath309 ( -50,2712)(0,0)[l]for @xmath279 outside the ( -50,2412)(0,0)[l]sectors ( as a site in @xmath54 ) hence @xmath310 depending on the position of the site @xmath279 .
in particular , this proves the formula ( [ recursivef ] ) for the boundary values .
let us continue with the proof of the first part of theorem [ zerocells ] .
from ( [ numberofsites ] ) we have @xmath311 .
remark [ exhaustx ] say that for each @xmath312 there exists @xmath313 such that the site @xmath314 , and hence @xmath315 , which shows that @xmath316 .
it is clear that @xmath317 is an injective function , for instance because @xmath318 determines @xmath5 as in ( [ recoverlambda ] ) .
hence we exhibited @xmath319 @xmath45-faces in @xmath200 , which is the maximal possible number of @xmath45-faces in the injective hull of a metric space with @xmath2 points ( the @xmath320 case of the following theorem [ hjthm ] by herrmann and joswig ) .
hence the @xmath45-faces of @xmath200 are precisely @xmath321 .
here is an example that illustrates the formula ( [ numberofsites ] ) with @xmath116 for the partition @xmath117 . each site in @xmath54
is represented thrice in the picture ( in other words , it shows three fundamental domains in the universal cover ) .
the displayed numbers are the values of @xmath290 .
( the dashed rectangle will serve as an illustration of remark [ inaffhypplane ] . ) ( 10824,3339)(0,-10 ) ( 12,3012)(312,3312)(612,3012 ) ( 912,3312)(1212,3012)(1512,3312 ) ( 1812,3012)(2112,3312)(2412,3012 ) ( 2712,3312)(3012,3012)(3312,3312 ) ( 3612,3012)(3912,3312)(4212,3012 ) ( 4512,3312)(4812,3012)(5112,3312 ) ( 5412,3012)(5712,3312)(6012,3012 ) ( 6312,3312)(6612,3012)(6912,3312 ) ( 7212,3012)(7512,3312)(7812,3012 ) ( 8112,3312)(8412,3012)(8712,3312 ) ( 9012,3012)(9312,3312)(9612,3012 ) ( 9912,3312)(10212,3012)(10512,3312 ) ( 10812,3012)(7812,12)(7512,312 ) ( 7212,12)(6912,312)(6612,12 ) ( 6312,312)(6012,12)(5712,312 ) ( 5412,12)(5112,312)(4812,12 ) ( 4512,312)(4212,12)(3912,312 ) ( 3612,12)(3312,312)(3012,12)(12,3012 ) ( 1512,1512)(2412,2412)(2712,2112 ) ( 3012,2412)(3612,1812)(4212,2412 ) ( 5112,1512)(5412,1812)(5712,1512 ) ( 6312,2112)(6912,1512)(7812,2412 ) ( 8112,2112)(8412,2412)(9012,1812 ) ( 9612,2412)(9912,2112)(9012,1212 ) ( 8412,1812)(8112,1512)(7812,1812 ) ( 6912,912)(6312,1512)(5712,912 ) ( 5412,1212)(5112,912)(4212,1812 ) ( 3612,1212)(3012,1812)(2712,1512 ) ( 2412,1812)(1812,1212)(1512,1512 ) ( 1512,1512)(2412,2412)(2712,2112 ) ( 3012,2412)(3612,1812)(4212,2412 ) ( 5112,1512)(5412,1812)(5712,1512 ) ( 6312,2112)(6912,1512)(7812,2412 ) ( 8112,2112)(8412,2412)(9012,1812 ) ( 9612,2412)(9912,2112)(9012,1212 ) ( 8412,1812)(8112,1512)(7812,1812 ) ( 6912,912)(6312,1512)(5712,912 ) ( 5412,1212)(5112,912)(4212,1812 ) ( 3612,1212)(3012,1812)(2712,1512 ) ( 2412,1812)(1812,1212)(1512,1512 ) ( 1812,1212)(3012,12)(4512,1512)(4212,1812)(3612,1212)(3012,1812)(2712,1512 ) ( 2412,1812)(1812,1212 ) ( 3912,2112)(5112,3312)(7212,1212)(6012,12)(3912,2112 ) ( 1812,1512)(0,0)@xmath45 ( 4512,1812)(0,0)@xmath45 ( 7212,1512)(0,0)@xmath45 ( 2112,1212)(0,0)@xmath48 ( 2412,912)(0,0)@xmath8 ( 2712,612)(0,0)@xmath133 ( 3012,312)(0,0)@xmath322 ( 2412,1512)(0,0)@xmath4 ( 2712,1212)(0,0)@xmath49 ( 3012,912)(0,0)@xmath8 ( 3312,612)(0,0)@xmath323 ( 3612,312)(0,0)@xmath322 ( 3012,1512)(0,0)@xmath4 ( 3312,1212)(0,0)@xmath48 ( 3612,912)(0,0)@xmath8 ( 3912,612)(0,0)@xmath133 ( 4212,312)(0,0)@xmath324 ( 3912,1212)(0,0)@xmath48 ( 4212,912)(0,0)@xmath47 ( 4512,612)(0,0)@xmath135 ( 4812,312)(0,0)@xmath323 ( 4212,1512)(0,0)@xmath4 ( 4512,1212)(0,0)@xmath48 ( 4812,912)(0,0)@xmath49 ( 5112,612)(0,0)@xmath8 ( 5412,312)(0,0)@xmath323 ( 5412,912)(0,0)@xmath4 ( 5712,612)(0,0)@xmath47 ( 6012,312)(0,0)@xmath134 ( 6012,912)(0,0)@xmath48 ( 6312,612)(0,0)@xmath47 ( 6612,312)(0,0)@xmath323 ( 6312,1212)(0,0)@xmath4 ( 6612,912)(0,0)@xmath48 ( 6912,612)(0,0)@xmath135 ( 7212,312)(0,0)@xmath324 ( 7212,912)(0,0)@xmath49 ( 7512,612)(0,0)@xmath134 ( 7812,312)(0,0)@xmath324 ( 7512,1212)(0,0)@xmath48 ( 7812,912)(0,0)@xmath8 ( 8112,612)(0,0)@xmath133 ( 7812,1512)(0,0)@xmath4 ( 8112,1212)(0,0)@xmath49 ( 8412,912)(0,0)@xmath8 ( 8412,1512)(0,0)@xmath4 ( 8712,1212)(0,0)@xmath48 ( 4812,2112)(0,0)@xmath48 ( 5112,2412)(0,0)@xmath8 ( 5412,2712)(0,0)@xmath133 ( 5712,3012)(0,0)@xmath322 ( 6012,2712)(0,0)@xmath323 ( 6312,2412)(0,0)@xmath8 ( 6612,2112)(0,0)@xmath48 ( 6912,1812)(0,0)@xmath4 ( 6012,2112)(0,0)@xmath48 ( 5712,2412)(0,0)@xmath8 ( 5112,3012)(0,0)@xmath324 ( 4812,2712)(0,0)@xmath134 ( 4512,2412)(0,0)@xmath49 ( 5112,1812)(0,0)@xmath4 ( 5712,1812)(0,0)@xmath4 ( 5412,2112)(0,0)@xmath49 [ inaffhypplane ] for @xmath325 ( with @xmath247 ) the formula @xmath326 from theorem [ zerocells ] ( see also definition [ hldjk ] ) has the following interpretation : the shapes of @xmath327 and @xmath328 together with the rim @xmath108 emanating from the position @xmath56 and ending at the position @xmath329 as a site in tessellate the @xmath330@xmath90@xmath331 rectangle with corners @xmath56 , @xmath332 , @xmath329 , @xmath301 . hence @xmath333 as an illustration we look at the example above and take @xmath334 .
the picture shows the corresponding dashed @xmath335 rectangle .
so @xmath336 the formula ( [ complement ] ) in section [ continuous ] is a continuous analogue of ( [ complementdiscrete ] ) .
[ hjthm ] the number of @xmath337-faces in an injective hull of a metric space with @xmath2 points is at most @xmath338 and for each @xmath2 there is a metric space attaining those upper bounds uniformly for all @xmath337 .
let @xmath275 with outer rim @xmath108 as in proposition [ loopsandpartitions ] .
let us enumerate those sites @xmath339 in the outer rim @xmath108 that correspond to the inner corners @xmath340 of @xmath5 .
the partition @xmath5 covers exactly @xmath341 partitions , namely those partitions @xmath6 that are got by removing from @xmath5 one of its inner corners .
the @xmath342 partitions @xmath343 that are got from @xmath5 by removing @xmath344 ( running from @xmath45 to @xmath341 ) of its inner corners constitute a boolean lattice .
we shall recognize that the convex hull of those @xmath342 vertices @xmath345 belongs to the injective hull @xmath200 .
in fact , let @xmath346 be such a convex combination . then from @xmath347 for all @xmath245 we have @xmath348 for all @xmath245 , that is , @xmath349 .
for each @xmath350 we have @xmath351 and hence @xmath352 .
since @xmath353 , we still get ( as a generalization of remark [ exhaustx ] ) @xmath354 from which we conclude that @xmath355 .
( note also that from ( [ exhaustxgen ] ) we get @xmath356 .
this is develin s bound for the maximal possible dimension of the injective hull of a metric space with @xmath2 points ( see @xcite ) , a result that is superseded by theorem [ hjthm ] . )
a systematic count of @xmath337-faces can be done as follows : each choice of @xmath337 among the @xmath341 inner corners of @xmath5 results in a @xmath337-face with @xmath259 as its `` top vertex '' . in this way
, we get @xmath357 @xmath337-faces with `` top vertex '' @xmath259 .
the total number of @xmath337-faces that we get from this procedure ( and we shall soon see that every @xmath337-face occurs in such a way ) is thus @xmath358 where @xmath359 denotes the number of inner corners of the partition @xmath5 . the number of partitions with first part at most @xmath42 and having at most @xmath131 parts that cover exactly @xmath341 partitions is @xmath360 .
we have to count the partitions with first part at most @xmath42 and at most @xmath131 parts with exactly @xmath341 inner corners .
this amounts to choosing @xmath341 integers @xmath361 and @xmath341 integers @xmath362 because the pairs @xmath363 for @xmath364 determine the positions of the @xmath341 inner corners of a partition in the given range in an evident way as illustrated in the picture .
( 6324,5739)(0,-10 ) ( 2112,2712 ) ( 2112,2712 ) ( 2712,2712 ) ( 2712,2712 ) ( 4212,3012 ) ( 4212,3012 ) ( 5112,3312 ) ( 5112,3312 ) ( 1812,2712)(2112,2412)(2412,2712 ) ( 2412,2712)(2712,2412)(3012,2712 ) ( 3912,3012)(4212,2712)(4512,3012 ) ( 4812,3312)(5112,3012 ) ( 1212,2112)(2112,3012)(2412,2712 ) ( 2712,3012)(3312,2412)(4212,3312 ) ( 4512,3012)(5112,3612)(5412,3312 ) ( 2712,612)(1212,2112 ) ( 1212,2112)(612,2712 ) ( 1212,2112)(612,2712 ) ( 5412,3312)(5712,3612 ) ( 5412,3312)(5712,3612 ) ( 118.066,2048.360)(12.000,2112.000)(75.640,2005.934 ) ( 12,2112)(2112,12 ) ( 12,2112)(2112,12 ) ( 2005.934,75.640)(2112.000,12.000)(2048.360,118.066 ) ( 3375.640,118.066)(3312.000,12.000)(3418.066,75.640 ) ( 3312,12)(6312,3012 ) ( 3312,12)(6312,3012 ) ( 6248.360,2905.934)(6312.000,3012.000)(6205.934,2948.360 ) ( 612,2712)(3612,5712)(5712,3612 ) ( 1512,1812)(2112,2412)(3312,1212 ) ( 1812,1512)(2712,2412)(3612,1512 ) ( 2112,1212)(3312,2412)(3912,1812 ) ( 2412,912)(4212,2712)(4512,2412 ) ( 3612,2712)(4212,2112 ) ( 4512,3012)(4812,2712 ) ( 1512,2412)(3012,912 ) ( 3612,1212)(0,0)@xmath365 ( 3912,1512)(0,0)@xmath366 ( 4812,2412)(0,0)@xmath367 ( 5412,3012)(0,0)@xmath368 ( 2412,612)(0,0)@xmath369 ( 2112,912)(0,0)@xmath370 ( 1512,1512)(0,0)@xmath371 ( 1212,1812)(0,0)@xmath372 ( 5112,1212)(0,0)@xmath42 ( 612,612)(0,0)@xmath131 the correspondence is clearly bijective .
[ coveriny ] the number of partitions in @xmath1 that cover exactly @xmath341 partitions ( that is , with exactly @xmath341 inner corners ) is @xmath373 .
let us compute the number in question by inclusion - exclusion and using the previous lemma and its notation @xmath374 for the number of partitions with first part at most @xmath42 and having at most @xmath131 parts that cover exactly @xmath341 partitions .
the formula is obviously true for @xmath375 ( corresponding to the empty partition , which is contained in every @xmath1 ) ; hence we assume that @xmath376 .
the following figure explains the inductive step for the inclusion - exclusion argument .
( 8124,5739)(0,-10 ) ( 118.066,2048.360)(12.000,2112.000)(75.640,2005.934 ) ( 12,2112)(2112,12 ) ( 12,2112)(2112,12 ) ( 2005.934,75.640)(2112.000,12.000)(2048.360,118.066 ) ( 3375.640,118.066)(3312.000,12.000)(3418.066,75.640 ) ( 3312,12)(6312,3012 ) ( 3312,12)(6312,3012 ) ( 6248.360,2905.934)(6312.000,3012.000)(6205.934,2948.360 ) ( 2712,612)(7512,5412)(7212,5712 ) ( 6912,5412)(6612,5712)(6312,5412 ) ( 6012,5712)(5712,5412)(5412,5712 ) ( 5112,5412)(4812,5712)(4512,5412 ) ( 4212,5712)(912,2412)(2712,612 ) ( 2712,612)(5712,3612)(3612,5712 ) ( 612,2712)(2712,612 ) ( 6375.640,3118.066)(6312.000,3012.000)(6418.066,3075.640 ) ( 6312,3012)(8112,4812 ) ( 8048.360,4705.934)(8112.000,4812.000)(8005.934,4748.360 ) ( 5112,1212)(0,0)[l]@xmath217 ( 612,612)(0,0)@xmath131 ( 7512,3612)(0,0)[l]@xmath377 the number @xmath378 of partitions in @xmath1 having at most @xmath131 parts that cover exactly @xmath341 partitions is @xmath379 furthermore , @xmath380 .
the number @xmath381 that we want to compute is therefore @xmath382 where the last equality is clear from the following combinatorial interpretation : to choose @xmath383 integers @xmath384 is tantamount to first fixing @xmath385 and then choosing @xmath386 integers in @xmath387 and choosing @xmath341 integers in @xmath388 .
[ numberfaces ] the number of @xmath337-faces in @xmath37 is ( at least and in fact exactly ) @xmath389 this is clear from proposition [ coveriny ] together with ( [ vfacesine ] ) . [ allfaces ] consider the @xmath2-point metric space @xmath390 with metric @xmath212 .
the number of @xmath337-faces in its injective hull @xmath37 ( as in definition [ injhullx ] ) is @xmath391 for @xmath320 we have already proved this in theorem [ zerocells ] because @xmath392 .
hence we assume that @xmath393 .
using lemma [ numberfaces ] and three instances of the binomial series expansion ( as a formal power series ) @xmath394 ( namely , twice in the direction from left to right and once in the opposite direction ) , we compute the generating series @xmath395 & \stackrel{\phantom{(\ast)}}{=}\sum_{n\geqslant2 } \sum_{s = v}^{\lfloor n/2\rfloor}\binom{n}{2s}\binom sv q^n = \sum_{s = v}^\infty\sum_{n=2s}^\infty\binom{n}{2s}\binom sv q^n = \sum_{s = v}^\infty\binom sv\sum_{n=2s}^\infty\binom{n}{2s}q^n\\ & \stackrel{(\ast)}{= } \sum_{s = v}^\infty\binom sv\frac{q^{2s}}{(1-q)^{2s+1}}=\frac{1}{(1-q ) } \sum_{s = v}^\infty\binom sv\left(\frac{q^2}{(1-q)^2}\right)^s\\ & \stackrel{(\ast)}{= } \frac{1}{(1-q)}\frac{\left(\dfrac{q^2}{(1-q)^2}\right)^v } { \left(1-\dfrac{q^2}{(1-q)^2}\right)^{v+1 } } = \frac{(1-q)q^{2v}}{(1 - 2q)^{v+1}}=(1-q)q^v 2^{-v}\frac{(2q)^v}{(1 - 2q)^{v+1}}\\ & \stackrel{(\ast)}{=}(1-q)q^v 2^{-v}\sum_{n = v}^\infty\binom nv ( 2q)^n\\ & \stackrel{\phantom{(\ast)}}{=}q^v 2^{-v-1}\sum_{n=2v}^\infty2\binom{n - v}{v}(2q)^{n - v } -q^{v+1 } 2^{-v}\sum_{n=2v+1}^\infty\binom{n - v-1}{v}(2q)^{n - v-1}\\ \intertext{and using $ 2\binom{n - v}{v}-\binom{n - v-1}{v}= \frac{n}{n - v}\binom{n - v}{v}$ and $ \binom{n - v-1}{v}\bigr|_{n=2v}=0 $ , we get finally } & \stackrel{\phantom{(\ast)}}{= } \sum_{n=2v}^\infty 2^{n-2v-1}\frac{n}{n - v}\binom{n - v}{v}\,q^n.\end{aligned}\ ] ] hence the formula ( [ numbervfaces ] ) is proved . according to theorem [ hjthm ]
, @xmath34 is a metric space such that @xmath37 attains the maximal possible number of @xmath337-faces for all @xmath337 .
[ proofsummarythm ] theorem [ summarythm ] from the introductory section follows as a corollary .
in fact , the vertices of @xmath37 were described in theorem [ zerocells ] as @xmath396 for @xmath397 .
the considerations after theorem [ hjthm ] showed that if @xmath398 covers @xmath6 , then the segment with endpoints @xmath259 and @xmath399 is an edge in @xmath37 . in this way
we get @xmath400 edges in @xmath37 , and according to theorem [ allfaces ] , those @xmath400 edges are all the @xmath4-faces in @xmath37 . for @xmath401
we get @xmath400 edges in @xmath37 and thus in @xmath7 .
this result generalizes to the fact that the hasse diagram of the poset of abelian ideals in a borel subalgebra of a complex simple lie algebra of rank @xmath402 has @xmath403 edges ( see ( * ? ? ?
* theorem 4.1 ) ) . for @xmath404 such a simple ( only rank - dependent ) census breaks down .
recall that @xmath405 ( type @xmath406 ) has five @xmath48-faces and no higher - dimensional faces .
but the poset of abelian ideals in a borel subalgebra of @xmath407 ( type @xmath408 ) contains a boolean subposet of rank @xmath49 : if @xmath409 are the simple roots and @xmath410 is the highest root , then the eight ideals in question are @xmath411 where @xmath16 denotes the root space for the root @xmath17 .
hence the hasse diagram contains the @xmath4-skeleton of a cube , which has one @xmath49-face and six @xmath48-faces .
its vertices @xmath412 are encoded by the graphs with vertex set @xmath413 , and there is an edge between @xmath42 and @xmath131 if and only if @xmath325 ( the outer rim of @xmath5 as defined in proposition [ loopsandpartitions ] ) .
@r=12mm@c=1 mm & & & ( 3991,4089)(0,-10 ) ( 2439,3754 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 2347,3469)(3383,2042 ) ( 2347,3469)(669,1160 ) ( 2347,3469)(2347,615 ) ( 2347,3469)(669,2924 ) & & ( 3991,4089)(0,-10 ) ( 2439,330 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 2347,615)(3383,2042 ) ( 2347,615)(669,1160 ) ( 2347,615)(2347,3469 ) ( 2347,615)(669,2924 ) + & & & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 2347,3469)(3383,2042 ) ( 2347,3469)(669,1160 ) ( 2347,3469)(2347,615 ) ( 2347,3469)(669,2924 ) ( 669,1160)(2347,615 ) @-[u ] & & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 2347,615)(3383,2042 ) ( 2347,615)(669,1160 ) ( 2347,615)(2347,3469 ) ( 2347,615)(669,2924 ) ( 2347,3469)(669,2924 ) @-[u ] + ( 3991,4089)(0,-10 ) ( 427,984 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 669,1160)(3383,2042 ) ( 669,1160)(2347,3469 ) ( 669,1160)(2347,615 ) ( 669,1160)(669,2924 ) & & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(2347,3469 ) ( 669,1160)(2347,3469 ) ( 669,1160)(2347,615 ) ( 669,1160)(669,2924 ) ( 2347,3469)(669,2924 ) @-[ur ] & [ 0pt][0pt ] & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(2347,615 ) ( 669,1160)(2347,615 ) ( 2347,3469)(2347,615 ) ( 2347,3469)(669,2924 ) ( 3383,2042)(2347,3469 ) @-[ul]@-[ur ] & [ 0pt][0pt ] & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(2347,615 ) ( 669,1160)(2347,615 ) ( 669,1160)(669,2924 ) ( 2347,3469)(669,2924 ) ( 2347,615)(669,2924 ) @-[ul ] & & ( 3991,4089)(0,-10 ) ( 427,3100 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 669,2924)(3383,2042 ) ( 669,2924)(669,1160 ) ( 669,2924)(2347,3469 ) ( 669,2924)(2347,615 ) + & & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(2347,3469 ) ( 669,1160)(2347,3469 ) ( 669,1160)(669,2924 ) ( 669,1160)(2347,615 ) ( 3383,2042)(669,1160 ) @-[ull]@-[u ] & [ 0pt][0pt ] & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(2347,615 ) ( 669,1160)(2347,615 ) ( 669,1160)(669,2924 ) ( 2347,3469)(669,2924 ) ( 3383,2042)(2347,3469 ) @-[ull ] @-[u]@-[urr ] & [ 0pt][0pt ] & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(669,2924 ) ( 669,1160)(669,2924 ) ( 2347,3469)(669,2924 ) ( 2347,615)(669,2924 ) ( 3383,2042)(2347,615 ) @-[u]@-[urr ] + & & & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(2347,615 ) ( 669,1160)(2347,615 ) ( 669,1160)(669,2924 ) ( 3383,2042)(669,1160 ) ( 3383,2042)(2347,3469 ) @-[ul]@-[ur ] & [ 0pt][0pt ] & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(669,2924 ) ( 669,1160)(669,2924 ) ( 2347,3469)(669,2924 ) ( 3383,2042)(2347,3469 ) ( 3383,2042)(2347,615 ) @-[ul]@-[ur ] + & & & & ( 3991,4089)(0,-10 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(669,2924 ) ( 669,1160)(669,2924 ) ( 3383,2042)(669,1160 ) ( 3383,2042)(2347,3469 ) ( 3383,2042)(2347,615 ) @-[ul]@-[ur ] + & & & & ( 3991,4089)(0,-10 ) ( 3683,2042 ) ( 3683,2042)(0,0)@xmath45 ( 427,984)(0,0)@xmath4 ( 2439,3754)(0,0)@xmath48 ( 2439,330)(0,0)@xmath49 ( 427,3100)(0,0)@xmath47 ( 3383,2042)(669,1160 ) ( 3383,2042)(2347,3469 ) ( 3383,2042)(2347,615 ) ( 3383,2042)(669,2924 ) @-[u ] the five graphs in the framed boxes encode the @xmath48-faces of @xmath405 in an evident manner ( and the twenty edges of @xmath414 or @xmath4-faces of @xmath405 are encoded by the obvious ( but not separately displayed ) spanning trees ) . the vertices in the first visualization show the outer rims ( as defined in proposition [ loopsandpartitions ] ) .
the second visualization shows the graphs on @xmath415 that encode the outer rims .
the points @xmath416 correspond to the rectangular partitions @xmath417 , where @xmath418 is synonymous with the empty partition .
recall that @xmath419 the @xmath2@xmath90@xmath2 symmetric circulant matrix @xmath420 is invertible ( for @xmath421 ) ; in fact , its determinant is @xmath422 where the first equality follows by writing @xmath423 as the product of the eigenvalues of @xmath424 ( by using the obvious eigenvectors ) and the rest is computation ( given without details ) .
we use the matrix @xmath425 to project @xmath426 to the plane , so that @xmath427 are the vertices of a regular @xmath2-gon .
the following picture shows the image of the @xmath4-skeleton of @xmath200 for @xmath116 .
we have @xmath428 for the fixed point @xmath429 under the action of the cyclic group generated by @xmath269 ( see theorem [ zerocells ] ) , but also for @xmath430 , which is an orbit of size three , we have @xmath428 .
all the other orbits have size nine .
let us consider as a continuous version of integer partitions certain functions @xmath431&\longrightarrow\mathbb
r_{\geqslant0}\\\notag u&\longmapsto -u\\\notag v&\longmapsto v\end{aligned}\ ] ] with @xmath432 such that the set @xmath433,\ |t|\leqslant s\leqslant\lambda(t)\bigr\ } \subseteq\mathbb r^2 $ ] ( 6168,3480)(0,-10 ) ( 15,3315)(3015,315)(6015,3315 ) ( 15,315)(6015,315 ) ( 1815,315)(1815,1515 ) ( 4440,315)(4440,1740 ) ( 1815,1515)(1816,1514)(1819,1512 ) ( 1823,1508)(1830,1502)(1838,1496 ) ( 1849,1489)(1863,1482)(1880,1473 ) ( 1902,1463)(1930,1452)(1965,1440 ) ( 1993,1431)(2020,1422)(2043,1415 ) ( 2060,1408)(2072,1403)(2080,1398 ) ( 2086,1394)(2090,1390)(2095,1386 ) ( 2101,1383)(2112,1379)(2128,1376 ) ( 2150,1372)(2181,1369)(2220,1366 ) ( 2265,1365)(2303,1366)(2339,1367 ) ( 2372,1370)(2400,1373)(2423,1376 ) ( 2443,1378)(2458,1381)(2470,1384 ) ( 2481,1387)(2490,1390)(2499,1393 ) ( 2510,1396)(2522,1400)(2537,1404 ) ( 2557,1408)(2580,1413)(2608,1419 ) ( 2641,1425)(2677,1433)(2715,1440 ) ( 2760,1449)(2798,1458)(2829,1465 ) ( 2851,1470)(2866,1475)(2875,1478 ) ( 2880,1481)(2884,1484)(2887,1486 ) ( 2891,1489)(2899,1492)(2911,1495 ) ( 2929,1500)(2953,1505)(2982,1510 ) ( 3015,1515)(3051,1520)(3080,1523 ) ( 3102,1525)(3116,1526)(3123,1527 ) ( 3127,1527)(3128,1527)(3132,1527 ) ( 3139,1526)(3153,1525)(3175,1523 ) ( 3204,1520)(3240,1515)(3273,1510 ) ( 3303,1504)(3327,1498)(3347,1492 ) ( 3361,1487)(3371,1481)(3378,1476 ) ( 3384,1471)(3390,1466)(3397,1461 ) ( 3407,1457)(3423,1452)(3443,1448 ) ( 3470,1444)(3503,1441)(3540,1440 ) ( 3577,1441)(3609,1443)(3635,1445 ) ( 3653,1448)(3665,1449)(3672,1451 ) ( 3676,1452)(3679,1454 ) ( 3684,1456)(3692,1459)(3705,1465 ) ( 3727,1472)(3757,1483)(3795,1498 ) ( 3840,1515)(3874,1529)(3908,1543 ) ( 3939,1557)(3966,1570)(3991,1582 ) ( 4011,1593)(4029,1603)(4043,1611 ) ( 4056,1619)(4067,1627)(4078,1634 ) ( 4088,1641)(4099,1648)(4111,1656 ) ( 4125,1664)(4141,1673)(4160,1683 ) ( 4182,1694)(4206,1705)(4233,1717 ) ( 4262,1729)(4290,1740)(4312,1748 ) ( 4332,1755)(4350,1760)(4365,1764 ) ( 4378,1766)(4389,1768)(4399,1769 ) ( 4407,1769)(4440,1740 ) ( 3015,315)(1815,1515)(1816,1514)(1819,1512 ) ( 1823,1508)(1830,1502)(1838,1496 ) ( 1849,1489)(1863,1482)(1880,1473 ) ( 1902,1463)(1930,1452)(1965,1440 ) ( 1993,1431)(2020,1422)(2043,1415 ) ( 2060,1408)(2072,1403)(2080,1398 ) ( 2086,1394)(2090,1390)(2095,1386 ) ( 2101,1383)(2112,1379)(2128,1376 ) ( 2150,1372)(2181,1369)(2220,1366 ) ( 2265,1365)(2303,1366)(2339,1367 ) ( 2372,1370)(2400,1373)(2423,1376 ) ( 2443,1378)(2458,1381)(2470,1384 ) ( 2481,1387)(2490,1390)(2499,1393 ) ( 2510,1396)(2522,1400)(2537,1404 ) ( 2557,1408)(2580,1413)(2608,1419 ) ( 2641,1425)(2677,1433)(2715,1440 ) ( 2760,1449)(2798,1458)(2829,1465 ) ( 2851,1470)(2866,1475)(2875,1478 ) ( 2880,1481)(2884,1484)(2887,1486 ) ( 2891,1489)(2899,1492)(2911,1495 ) ( 2929,1500)(2953,1505)(2982,1510 ) ( 3015,1515)(3051,1520)(3080,1523 ) ( 3102,1525)(3116,1526)(3123,1527 ) ( 3127,1527)(3128,1527)(3132,1527 ) ( 3139,1526)(3153,1525)(3175,1523 ) ( 3204,1520)(3240,1515)(3273,1510 ) ( 3303,1504)(3327,1498)(3347,1492 ) ( 3361,1487)(3371,1481)(3378,1476 ) ( 3384,1471)(3390,1466)(3397,1461 ) ( 3407,1457)(3423,1452)(3443,1448 ) ( 3470,1444)(3503,1441)(3540,1440 ) ( 3577,1441)(3609,1443)(3635,1445 ) ( 3653,1448)(3665,1449)(3672,1451 ) ( 3676,1452)(3679,1454 ) ( 3684,1456)(3692,1459)(3705,1465 ) ( 3727,1472)(3757,1483)(3795,1498 ) ( 3840,1515)(3874,1529)(3908,1543 ) ( 3939,1557)(3966,1570)(3991,1582 ) ( 4011,1593)(4029,1603)(4043,1611 ) ( 4056,1619)(4067,1627)(4078,1634 ) ( 4088,1641)(4099,1648)(4111,1656 ) ( 4125,1664)(4141,1673)(4160,1683 ) ( 4182,1694)(4206,1705)(4233,1717 ) ( 4262,1729)(4290,1740)(4312,1748 ) ( 4332,1755)(4350,1760)(4365,1764 ) ( 4378,1766)(4389,1768)(4399,1769 ) ( 4407,1769)(4440,1740)(3015,315 ) ( 3015,315)(3015,3315 ) ( 15,270)(3000,0)3(0,1)90 ( 2885,3315)(0,0)[r]@xmath4 ( 2970,3315)(1,0)90 ( 15,-15)(0,0)[b]@xmath434 ( 6015,-15)(0,0)[b]@xmath4 ( 3015,-15)(0,0)[b]@xmath45 ( 1815,-15)(0,0)[b]@xmath435 ( 4440,-15)(0,0)[b]@xmath337 has an evident meaning of `` continuous young diagram '' of the `` continuous partition '' @xmath436 . that the parts of an integer partition are ordered decreasingly translates into the requirement that @xmath436 must be @xmath4-lipschitz .
we further require that @xmath436 be defined on an interval @xmath437 $ ] of length at most @xmath4 ( which replaces the bound on the hook lengths for integer partitions in @xmath1 ) .
there is then a unique extension of @xmath436 to a @xmath4-lipschitz function @xmath438\to\mathbb r_{\geqslant0}$ ] with @xmath439 , namely , @xmath440 for @xmath441 $ ] .
note that @xmath442 for all @xmath443 $ ] because for @xmath444 the @xmath4-lipschitz condition gives @xmath445 and hence @xmath446 the shifted antiperiodic extension defined by @xmath447 finally extends @xmath436 to the real line @xmath448 $ ] , so that the graph of @xmath436 becomes the preimage of a loop in the mbius strip under the universal covering projection .
we define @xmath449\bigm| \mbox{$\lambda$ is $ 1$-lipschitz and $ \lambda(t+1)=1-\lambda(t)$ for all $ t$}\bigr\}.\ ] ] for each
@xmath450 its graph intersects the line segment with endpoints @xmath451 and @xmath64 in at least one ( and generically exactly one ) point . if @xmath452 is a point in the intersection , then the restriction @xmath453}$ ] defines a `` continuous partition '' as in ( [ continuouspartition ] ) with @xmath454 .
( 13824,3680)(0,-10 ) ( 1812,2115)(3012,3315)(4318,2009 ) ( 6012,315)(7318,1621 ) ( 7318,1621)(9012,3315 ) ( 9012,3315)(10812,1515 ) ( 9012,1515)(9014,1515)(9019,1515 ) ( 9028,1516)(9042,1517)(9060,1518 ) ( 9084,1520)(9112,1522)(9144,1525 ) ( 9179,1528)(9215,1532)(9253,1536 ) ( 9291,1541)(9328,1546)(9364,1552 ) ( 9400,1558)(9434,1565)(9467,1573 ) ( 9498,1581)(9529,1590)(9560,1600 ) ( 9590,1611)(9620,1623)(9650,1636 ) ( 9681,1650)(9712,1665)(9737,1677 ) ( 9762,1691)(9787,1705)(9814,1719 ) ( 9841,1734)(9869,1750)(9898,1767 ) ( 9927,1784)(9958,1802)(9989,1821 ) ( 10021,1839)(10053,1859)(10087,1878 ) ( 10121,1898)(10155,1918)(10190,1938 ) ( 10226,1958)(10262,1978)(10297,1998 ) ( 10333,2017)(10369,2036)(10405,2054 ) ( 10441,2072)(10476,2089)(10511,2105 ) ( 10546,2121)(10580,2136)(10615,2150 ) ( 10648,2163)(10681,2175)(10714,2187 ) ( 10747,2197)(10780,2206)(10812,2215 ) ( 10844,2223)(10877,2229)(10910,2235 ) ( 10943,2240)(10976,2243)(11011,2246 ) ( 11047,2247)(11083,2248)(11122,2248 ) ( 11161,2246)(11203,2244)(11246,2241 ) ( 11291,2237)(11338,2232)(11387,2226 ) ( 11437,2219)(11489,2211)(11506,2209 ) ( 2506,1421)(2542,1427 ) ( 2595,1436)(2648,1445)(2699,1454 ) ( 2750,1463)(2797,1472)(2841,1480 ) ( 2880,1488)(2915,1495)(2944,1501 ) ( 2967,1505)(2985,1509)(2998,1512 ) ( 3006,1514)(3010,1515)(3012,1515 ) ( 3012,1515)(3014,1515)(3019,1515 ) ( 3028,1516)(3042,1517)(3060,1518 ) ( 3084,1520)(3112,1522)(3144,1525 ) ( 3179,1528)(3215,1532)(3253,1536 ) ( 3291,1541)(3328,1546)(3364,1552 ) ( 3400,1558)(3434,1565)(3467,1573 ) ( 3498,1581)(3529,1590)(3560,1600 ) ( 3590,1611)(3620,1623)(3650,1636 ) ( 3681,1650)(3712,1665)(3737,1677 ) ( 3762,1691)(3787,1705)(3814,1719 ) ( 3841,1734)(3869,1750)(3898,1767 ) ( 3927,1784)(3958,1802)(3989,1821 ) ( 4021,1839)(4053,1859)(4087,1878 ) ( 4121,1898)(4155,1918)(4190,1938 ) ( 4226,1958)(4262,1978)(4297,1998)(4318,2009 ) ( 4333,2017)(4369,2036)(4405,2054 ) ( 4441,2072)(4476,2089)(4511,2105 ) ( 4546,2121)(4580,2136)(4615,2150 ) ( 4648,2163)(4681,2175)(4714,2187 ) ( 4747,2197)(4780,2206)(4812,2215 ) ( 4844,2223)(4877,2229)(4910,2235 ) ( 4943,2240)(4976,2243)(5011,2246 ) ( 5047,2247)(5083,2248)(5122,2248 ) ( 5161,2246)(5203,2244)(5246,2241 ) ( 5291,2237)(5338,2232)(5387,2226 ) ( 5437,2219)(5489,2211)(5542,2203 ) ( 5595,2194)(5648,2185)(5699,2176 ) ( 5750,2167)(5797,2158)(5841,2150 ) ( 5880,2142)(5915,2135)(5944,2129 ) ( 5967,2125)(5985,2121)(5998,2118 ) ( 6006,2116)(6010,2115)(6012,2115 ) ( 6012,315)(4318,2009 ) ( 4333,2017)(4369,2036)(4405,2054 ) ( 4441,2072)(4476,2089)(4511,2105 ) ( 4546,2121)(4580,2136)(4615,2150 ) ( 4648,2163)(4681,2175)(4714,2187 ) ( 4747,2197)(4780,2206)(4812,2215 ) ( 4844,2223)(4877,2229)(4910,2235 ) ( 4943,2240)(4976,2243)(5011,2246 ) ( 5047,2247)(5083,2248)(5122,2248 ) ( 5161,2246)(5203,2244)(5246,2241 ) ( 5291,2237)(5338,2232)(5387,2226 ) ( 5437,2219)(5489,2211)(5542,2203 ) ( 5595,2194)(5648,2185)(5699,2176 ) ( 5750,2167)(5797,2158)(5841,2150 ) ( 5880,2142)(5915,2135)(5944,2129 ) ( 5967,2125)(5985,2121)(5998,2118 ) ( 6006,2116)(6010,2115)(6012,2115 ) ( 6014,2115)(6019,2115 ) ( 6028,2114)(6042,2113)(6060,2112 ) ( 6084,2110)(6112,2108)(6144,2105 ) ( 6179,2102)(6215,2098)(6253,2094 ) ( 6291,2089)(6328,2084)(6364,2078 ) ( 6400,2072)(6434,2065)(6467,2057 ) ( 6498,2049)(6529,2040)(6560,2030 ) ( 6590,2019)(6620,2007)(6650,1994 ) ( 6681,1980)(6712,1965)(6737,1953 ) ( 6762,1939)(6787,1925)(6814,1911 ) ( 6841,1896)(6869,1880)(6898,1863 ) ( 6927,1846)(6958,1828)(6989,1809 ) ( 7021,1791)(7053,1771)(7087,1752 ) ( 7121,1732)(7155,1712)(7190,1692 ) ( 7226,1672)(7262,1652)(7297,1632)(7318,1621 ) ( 6012,315 ) ( 6012,315)(4318,2009 ) ( 4333,2017)(4369,2036)(4405,2054 ) ( 4441,2072)(4476,2089)(4511,2105 ) ( 4546,2121)(4580,2136)(4615,2150 ) ( 4648,2163)(4681,2175)(4714,2187 ) ( 4747,2197)(4780,2206)(4812,2215 ) ( 4844,2223)(4877,2229)(4910,2235 ) ( 4943,2240)(4976,2243)(5011,2246 ) ( 5047,2247)(5083,2248)(5122,2248 ) ( 5161,2246)(5203,2244)(5246,2241 ) ( 5291,2237)(5338,2232)(5387,2226 ) ( 5437,2219)(5489,2211)(5542,2203 ) ( 5595,2194)(5648,2185)(5699,2176 ) ( 5750,2167)(5797,2158)(5841,2150 ) ( 5880,2142)(5915,2135)(5944,2129 ) ( 5967,2125)(5985,2121)(5998,2118 ) ( 6006,2116)(6010,2115)(6012,2115 ) ( 6014,2115)(6019,2115 ) ( 6028,2114)(6042,2113)(6060,2112 ) ( 6084,2110)(6112,2108)(6144,2105 ) ( 6179,2102)(6215,2098)(6253,2094 ) ( 6291,2089)(6328,2084)(6364,2078 ) ( 6400,2072)(6434,2065)(6467,2057 ) ( 6498,2049)(6529,2040)(6560,2030 ) ( 6590,2019)(6620,2007)(6650,1994 ) ( 6681,1980)(6712,1965)(6737,1953 ) ( 6762,1939)(6787,1925)(6814,1911 ) ( 6841,1896)(6869,1880)(6898,1863 ) ( 6927,1846)(6958,1828)(6989,1809 ) ( 7021,1791)(7053,1771)(7087,1752 ) ( 7121,1732)(7155,1712)(7190,1692 ) ( 7226,1672)(7262,1652)(7297,1632)(7318,1621 ) ( 6012,315 ) ( 6012,2115)(6014,2115)(6019,2115 ) ( 6028,2114)(6042,2113)(6060,2112 ) ( 6084,2110)(6112,2108)(6144,2105 ) ( 6179,2102)(6215,2098)(6253,2094 ) ( 6291,2089)(6328,2084)(6364,2078 ) ( 6400,2072)(6434,2065)(6467,2057 ) ( 6498,2049)(6529,2040)(6560,2030 ) ( 6590,2019)(6620,2007)(6650,1994 ) ( 6681,1980)(6712,1965)(6737,1953 ) ( 6762,1939)(6787,1925)(6814,1911 ) ( 6841,1896)(6869,1880)(6898,1863 ) ( 6927,1846)(6958,1828)(6989,1809 ) ( 7021,1791)(7053,1771)(7087,1752 ) ( 7121,1732)(7155,1712)(7190,1692 ) ( 7226,1672)(7262,1652)(7297,1632)(7318,1621 ) ( 7333,1613)(7369,1594)(7405,1576 ) ( 7441,1558)(7476,1541)(7511,1525 ) ( 7546,1509)(7580,1494)(7615,1480 ) ( 7648,1467)(7681,1455)(7714,1443 ) ( 7747,1433)(7780,1424)(7812,1415 ) ( 7844,1407)(7877,1401)(7910,1395 ) ( 7943,1390)(7976,1387)(8011,1384 ) ( 8047,1383)(8083,1382)(8122,1382 ) ( 8161,1384)(8203,1386)(8246,1389 ) ( 8291,1393)(8338,1398)(8387,1404 ) ( 8437,1411)(8489,1419)(8542,1427 ) ( 8595,1436)(8648,1445)(8699,1454 ) ( 8750,1463)(8797,1472)(8841,1480 ) ( 8880,1488)(8915,1495)(8944,1501 ) ( 8967,1505)(8985,1509)(8998,1512 ) ( 9006,1514)(9010,1515)(9012,1515 ) ( 6012,2115)(6012,315)(4318,2009)(4318,315)(7318,1621)(7318,315)(612,3315)(12612,3315 ) ( 2506,315 ) ( 9612,315 ) ( 2506,315)(2506,1421)(5506,315)(5506,2209)(612,3315)(3612,315)(6612,3315 ) ( 9612,315)(12612,3315 ) ( 6612,3385)(0,0)[b]@xmath65 ( 12612,3385)(0,0)[b]@xmath455 ( 612,3385)(0,0)[b]@xmath456 ( 3612,215)(0,0)[t]@xmath64 ( 9612,215)(0,0)[t]@xmath457 ( 6012,15)(0,0)[b]@xmath458 ( 5865,2215)(0,0)[lb]@xmath459 ( 4318,15)(0,0)[b]@xmath460 ( 7318,15)(0,0)[b]@xmath461[0pt]{$+$}1 $ ] ( 5506,15)(0,0)[b]@xmath462[0pt]{$+$}1 $ ] ( 2506,15)(0,0)[b]@xmath435 ( 4378,2309)(0,0)[b]@xmath463 ( 7300,1921)(0,0)[b]@xmath464 ( 2400,1320)(0,0)[r]@xmath465 ( 4318,2009)(5624,3315)(7318,1621)(8624,315)(10318,2009 ) ( 8624,15)(0,0)[b]@xmath341 ( 5623,3385)(0,0)[b]@xmath466 ( 11606,2209)(0,0)[l ] in analogy with the assignment @xmath467 we define @xmath468 by the ( @xmath48-periodic ) area function @xmath469 where @xmath470 and @xmath471 and hence @xmath472 ; moreover , @xmath473 .
+ for @xmath474 and hence @xmath475 we have @xmath476\notag f_\lambda(s)&=f_\lambda\bigl(b+\lambda(b)\bigr ) = f_\lambda\bigl(a+1+\lambda(a+1)\bigr)= f_\lambda\bigl(2+a-\lambda(a)\bigr)\phantom{\int_0 ^ 0}\\\notag & = \int_{a+1}^{u+2}\lambda(s)\,ds+\int_{u+2}^{a+2}\lambda(s)\,ds -\frac12\bigl(\lambda(a+1)^2+\lambda(a+2)^2\bigr)\\\notag & = \int_a^{u+1}\bigl(1-\lambda(s)\bigr)\,ds+\int_u^a\lambda(s)\,ds -\frac12\bigl((1-\lambda(a))^2+\lambda(a)^2\bigr)\\\label{fb } & = u+1-a+\int_u^a\lambda(s)\,ds-\int_a^{u+1}\lambda(s)\,ds + \lambda(a)\bigl(1-\lambda(a)\bigr)-\frac12\,.\end{aligned}\ ] ] the equality @xmath477 follows directly from the previous picture by considering the rectangle with corners at @xmath478 , @xmath479 , @xmath480 , @xmath481 ; alternatively , ( [ complement ] ) follows by adding ( [ fa ] ) and ( [ fb ] ) . for @xmath482 $ ] let @xmath483 be the function that satisfies @xmath484 ( 9134,3352)(0,-10 ) ( 912,315)(9012,315 ) ( 912,315)(912,2415 ) ( 3912,315)(3912,1215 ) ( 1812,315)(1815,3315 ) ( 912,2415)(1812,3315)(4812,315)(6912,2415 ) ( 6912,315)(6912,2415 ) ( 12,3315)(3012,315)(6012,3315)(9012,315 ) ( 12,3315)(6012,3315 ) ( 912,15)(0,0)[t]@xmath485 ( 3012,15)(0,0)[t]@xmath45 ( 3912,15)(0,0)[t]@xmath486 ( 4812,15)(0,0)[t]@xmath487 ( 1812,15)(0,0)[t]@xmath488 ( 6912,15)(0,0)[t]@xmath489 ( 9012,15)(0,0)[t]@xmath48 ( 7212,2415)(0,0)[l]graph of @xmath490 note that @xmath491 .
note also that @xmath492 is the unique function in @xmath493 that vanishes at @xmath487 .
the @xmath48-periodic function @xmath494 is then @xmath495 we define the metric space @xmath496 as the subspace @xmath497\bigr\}\subseteq\mathbb y_\infty.\ ] ] two alternative characterizations are @xmath498 the distance function is @xmath499 for @xmath500 $ ] the equality @xmath501 is an analogue of ( [ xembedex ] ) . @xmath502\to[0,1]\mbox { $ 1$-lipschitz}\\[-1 mm ] & \mbox{with $ \lambda(u)=-u$ and $ \lambda(u+1)=u+1$}\\[-1 mm ] & \mbox{extended by $ \lambda(t+1)=1-\lambda(t)$ to all $ t\in\mathbb r$}\\\hline ( 0,0),\,(0,n),\,(n , n);\,(j , k)&(-1,1),\,(0,0),\,(1,1);\ , \bigl(\frac jn+\frac kn-1,\frac jn-\frac kn+1\bigr)\\\hline ( j , k)\in\mathcal l_\lambda\mbox { ( for $ 0\leqslant j\leqslant k\leqslant
n$ ) } & \bigl(a,\lambda(a)\bigr)\mbox { ( for $ u\leqslant a \leqslant u+1$)}\\\hline d(j , k)=|k - j|\bigl(n-|k - j|\bigr)&d(r , s)=2|s - r|\bigl(1-|s - r|\bigr)\\\hline \forall\,(j , k)\in\mathcal l_\lambda:&\mbox{for $ b = a+1$:}\\[-1 mm ] f_\lambda(j)+f_\lambda(k)=d(j , k ) & f_\lambda\bigl(a+\lambda(a)\bigr)+f_\lambda\bigl(b+\lambda(b)\bigr ) = 2\lambda(a)\lambda(b)\\[-1 mm ] { } = d(0,|k - j|)&{}=d(0,\lambda(a))=d(0,\lambda(b))\\\hline x_n&x_\infty\\\hline \end{array}\ ] ] + of course one could elaborate on the correspondence by first putting the representation of integer partitions into a form analogous to the representation of `` continuous partitions '' . for instance , look at the partition @xmath503 with @xmath116 .
( 10824,4085)(0,-25 ) ( 12,3462)(312,3762)(612,3462 ) ( 912,3762)(1212,3462)(1512,3762 ) ( 1812,3462)(2112,3762)(2412,3462 ) ( 2712,3762)(3012,3462)(3312,3762 ) ( 3612,3462)(3912,3762)(4212,3462 ) ( 4512,3762)(4812,3462)(5112,3762 ) ( 5412,3462)(5712,3762)(6012,3462 ) ( 6312,3762)(6612,3462)(6912,3762 ) ( 7212,3462)(7512,3762)(7812,3462 ) ( 8112,3762)(8412,3462)(8712,3762 ) ( 9012,3462)(9312,3762)(9612,3462 ) ( 9912,3762)(10212,3462)(10512,3762 ) ( 10812,3462)(7812,462)(7512,762 ) ( 7212,462)(6912,762)(6612,462 ) ( 6312,762)(6012,462)(5712,762 ) ( 5412,462)(5112,762)(4812,462 ) ( 4512,762)(4212,462)(3912,762 ) ( 3612,462)(3312,762)(3012,462)(12,3462 ) ( 1512,1962)(2412,2862)(2712,2562 ) ( 3012,2862)(3612,2262)(4212,2862 ) ( 5112,1962)(5412,2262)(5712,1962 ) ( 6312,2562)(6912,1962)(7812,2862 ) ( 8112,2562)(8412,2862)(9012,2262 ) ( 9612,2862)(9912,2562)(9012,1662 ) ( 8412,2262)(8112,1962)(7812,2262 ) ( 6912,1362)(6312,1962)(5712,1362 ) ( 5412,1662)(5112,1362)(4212,2262 ) ( 3612,1662)(3012,2262)(2712,1962 ) ( 2412,2262)(1812,1662)(1512,1962 ) ( 1812,1662)(3012,462)(4512,1962 ) ( 4212,2262)(3612,1662)(3012,2262 ) ( 2712,1962)(2412,2262)(1812,1662 ) ( 1512,762)(10812,762 ) ( 10692.000,732.000)(10812.000,762.000)(10692.000,792.000 ) ( 3012,762)(3012,2412 ) ( 3012,2712)(3012,3762 ) ( 3042.000,3642.000)(3012.000,3762.000)(2982.000,3642.000 ) ( 2112,1362)(2712,1962 ) ( 2412,1062)(3312,1962 ) ( 2712,762)(3612,1662 ) ( 2112,1962)(3312,762 ) ( 2712,1962)(3612,1062 ) ( 3612,1662)(3912,1362 ) ( 3912,1962)(4212,1662 ) ( 1662,162)(1662,12)(4362,12 ) ( 4362,162)(4362,12)(7062,12)(7062,162 ) ( 4212,2262)(4512,2562 ) ( 6912,1962)(7212,1662 ) ( 1812,162)(0,0)@xmath504 ( 2112,162)(0,0)@xmath505 ( 2412,162)(0,0)@xmath506 ( 2712,162)(0,0)@xmath434 ( 3012,162)(0,0)@xmath45 ( 3312,162)(0,0)@xmath4 ( 3612,162)(0,0)@xmath48 ( 3912,162)(0,0)@xmath49 ( 4212,162)(0,0)@xmath47 ( 4512,162)(0,0)@xmath8 ( 4812,162)(0,0)@xmath135 ( 5112,162)(0,0)@xmath134 ( 5412,162)(0,0)@xmath133 ( 5712,162)(0,0)@xmath323 ( 6012,162)(0,0)@xmath9 ( 6312,162)(0,0)@xmath324 ( 6612,162)(0,0)@xmath507 ( 6912,162)(0,0)@xmath322 ( 7212,162)(0,0)@xmath508 ( 7512,162)(0,0)@xmath509 ( 7812,162)(0,0)@xmath510 ( 1812,1962)(0,0)@xmath47 ( 2112,2262)(0,0)@xmath8 ( 2412,2562)(0,0)@xmath135 ( 2712,2262)(0,0)@xmath8 ( 3012,2562)(0,0)@xmath135 ( 3312,2262)(0,0)@xmath8 ( 3612,1962)(0,0)@xmath47 ( 3912,2262)(0,0)@xmath8 ( 4212,2562)(0,0)@xmath135 ( 4512,2262)(0,0)@xmath8 ( 4812,1962)(0,0)@xmath47 ( 5112,1662)(0,0)@xmath49 ( 5412,1962)(0,0)@xmath47 ( 5712,1662)(0,0)@xmath49 ( 6012,1962)(0,0)@xmath47 ( 6312,2262)(0,0)@xmath8 ( 6612,1962)(0,0)@xmath47 ( 6912,1662)(0,0)@xmath49 ( 7212,1962)(0,0)@xmath47 ( 7512,2262)(0,0)@xmath8 ( 7812,2562)(0,0)@xmath135 ( 8112,2262)(0,0)@xmath8 ( 8412,2562)(0,0)@xmath135 ( 8712,2262)(0,0)@xmath8 ( 9012,1962)(0,0)@xmath47 ( 9312,2262)(0,0)@xmath8 ( 9612,2562)(0,0)@xmath135 ( 8112,162)(0,0)@xmath511 ( 8412,162)(0,0)@xmath512 ( 8712,162)(0,0)@xmath513 ( 9012,162)(0,0)@xmath514 ( 9312,162)(0,0)@xmath515 ( 9612,162)(0,0)@xmath516 the function @xmath517\cap\mathbb z\to[0,9]\cap\mathbb z$ ] @xmath518 describes the heights of the box positions in the outer rim of @xmath5 . in general , for @xmath275 we have the corresponding function @xmath519 defined on a set of @xmath2 consecutive integers that contains @xmath45 , and we extend it to a function @xmath520\cap\mathbb z$ ] by requiring the symmetry @xmath521 . in the example we have @xmath522 and @xmath523 , which is reminiscent of the boundary condition @xmath465 and
@xmath439 for a `` continuous partition '' @xmath438\to[0,1]$ ] .
let us also write down the formula @xmath524
the realization of @xmath7 as the @xmath4-skeleton of the @xmath525-dimensional polyhedral complex @xmath37 embedded in @xmath526 ( in fact , in its nonnegative orthant ) that is stable under cyclic permutation of the coordinates , provides a new geometric realization of the @xmath2-fold cyclic symmetry of @xmath7 .
when one looks at the graphs @xmath7 depicted in @xcite for @xmath527 , then the @xmath4-skeleta of five squares , one cube , seven cubes , respectively one tesseract are eye - catching . to some extent
, the construction of @xmath37 gives a meaning to those @xmath525-cubes .
in general , theorem [ allfaces ] counts all those as well as the lower - dimensional @xmath337-cubes .
the works of c. berg and m. zabrocki @xcite and h. thomas and n. williams @xcite generalize the cyclic symmetries and prove an instance of the cyclic sieving phenomenon in that framework .
one may ask whether the metric geometry approach developed here for the original cyclic symmetries can be extended in some nice way to the generalized version . homotopically nontrivial loops ( as specified in definition [ shortnontrivialloops ] ) play an important role .
an investigation of loops ( self - avoiding or not , of given length and with a prescribed winding number ) in various sorts of discrete strips ( also with other than @xmath15 type dynkin diagram fibres ) might involve interesting combinatorics .
let us conclude with some geometry .
the maximal coordinate sum of points in @xmath37 is attained for the @xmath2 points in the orbit of @xmath528 , @xmath529 the vertices of this @xmath530-cube are @xmath345 where @xmath343 is one of the @xmath531 partitions that are got by removing from the staircase partition @xmath532 any subset of its inner corners .
the barycentre of this @xmath530-cube has all coordinates @xmath533 , but no vertex of @xmath7 is fixed under the cyclic action ( which is also a very particular case of the cyclic sieving phenomenon established in @xcite ) .
recall that it is the supremum norm in @xmath526 that endows @xmath37 with the metric that characterizes this space as an injective hull .
but the polyhedral complex @xmath37 can also be looked at from a more geometric point of view by using the euclidean metric , so that all the edges have length @xmath534 by theorem [ zerocells ] . in the euclidean metric
the ` central cube ' for @xmath29 is actually a rhombohedron with six congruent faces , namely , rhombi with diagonals of lengths @xmath535 and @xmath47 ; one of the diagonals of the rhombohedron has length @xmath536 and the other three have length @xmath537 @xmath538which is @xmath539 . | a cascade of dihedral symmetries is hidden in young s lattice of integer partitions . in fact , for each @xmath0 the hasse graph of the subposet @xmath1 consisting of the partitions with maximal hook length strictly less than @xmath2 has the dihedral group of order @xmath3 as its symmetry group . here a new interpretation of those hasse graphs is presented , namely as the @xmath4-skeleta of the injective hulls of certain finite metric spaces . |
electrically and optically detected magnetic resonance experiments ( edmr and odmr , respectively ) are alternative ways to detect electron spin resonances ( esr ) in materials with charge carrier transport or recombination transitions that are governed by spin selection rules @xcite . the advantage of edmr and odmr in comparison to the traditional esr spectroscopy is the sensitivity of these methods which is typically 6 to 10 orders of magnitude higher @xcite .
this has become particularly useful for the investigation of paramagnetic centers in highly diluted matrices or low dimensional semiconductor thin film devices and interfaces , point defects @xcite and defect clusters .
one of the challenges of edmr and odmr spectroscopy is that the information obtained from these experiments is different in comparison to the esr data @xcite .
the reasons for the discrepancies between esr and edmr / odmr are mainly due to the two different measurement approaches which imply two different observables : when the density operator @xmath1 represents the spin ensemble to be investigated , the observable corresponding to esr experiments will always be spin polarization @xmath2 represented by the spin polarization operator @xmath3 whereas for the indirect detection through spin
dependent transport or recombination , the observables are the permutation symmetry or -antisymmetry operators represented by the singlet @xmath4 or triplet operators @xmath5 , respectively @xcite . for many experimental edmr / odmr studies ( the so called continuous wave experiments )
the different description of the observables is not relevant , since these experiments are carried out in the incoherent time regime where only a line shape analysis of the respective spectra is feasible .
however , when coherent effects are studied with pulsed techniques ( pedmr / podmr ) @xcite the interpretation of the experiments strongly relies on the proper theoretical description of spin interaction during coherent microwave excitation @xcite .
an example for the difference between a pedmr signal and an esr signal which come from the same spin ensemble are weakly exchange and weakly dipolar coupled distant pair states in the band gap of an arbitrary semiconductor material with weak spin orbit coupling as described analytically by boehme and lips @xcite
. weak spin
spin coupling means the coupling is much weaker than the difference @xmath6 of the larmor
frequencies @xmath7 of the pair partners @xmath8 and @xmath9 , respectively .
note that @xmath10 represents bohr s magneton and @xmath11 the magnitude of an external magnetic field to which the spin pair is exposed .
examples for such systems could be donor
acceptor pairs whose distance is sufficiently large , yet not large enough to make donor acceptor recombination impossible @xcite , or equivalently , trap dangling bond recombination in disordered silicon materials such as amorphous or microcrystalline silicon @xcite .
weak spin
orbit coupling is required in order to ensure spin conservation and therefore , a spin selection rule .
it is fulfilled for instance in all known silicon morphologies but also in many organic semiconductor materials @xcite .
when the two pair constituents are manipulated identically with a coherent pulse of high field strength @xmath12 ( @xmath13 , whereas @xmath14 is the gyromagnetic ratio ) , they undergo a simultaneous spin rabi oscillation .
this can be detected by means of pulse length dependence measurements with both , pesr and pedmr .
with pesr , the measurement could be conducted by integration of the free induction decay and would be called a transient nutation experiment @xcite whereas with pedmr , the rate relaxation after the coherent excitation would be integrated reflecting the pair permutation symmetry within the pairs at the end of the exciting pulse @xcite . while both , the pesr and the pedmr transients would exhibit oscillating signals , the frequency of these oscillations would differ by a factor of 2 : the pesr detected nutation frequency @xmath15 would simply represent the rabi frequency of an uncoupled spin @xmath16 , whereas the pedmr measured oscillation would exhibit the frequency at which the identically precessing spins of the two pair partners cross the geometric plane transverse to the field direction of the externally applied magnetic field @xmath11 ( the @xmath17-@xmath18 plane ) since at these moments , the projection of the parallel oriented spins in the @xmath17-@xmath18 plane onto the spin eigenstates with singlet content will be maximized . since this plane
is passed twice per nutation period , the oscillation of the transition rate is twice as high .
note that this frequency discrepancy of the oscillations between pesr and pedmr detected transient nutations is changed as the @xmath12field strength becomes weak : when @xmath19 , an esr excitation will be possible with only one pair partner at the same time .
thus , the maximum singlet content of a pair will be achieved when the spin orientations therein point into opposite directions parallel to the @xmath11 field .
hence , for weak @xmath12 fields , the nutation frequencies for pesr and pedmr become equal . in the following
, a semi analytical study is presented which describes spin dependent electronic transition rates ( e.g. recombination or hopping transport ) when the driving forces for rabi oscillation ( under experimental conditions these are typically strong coherent microwave fields ) are between a weak ( @xmath19 ) and a strong excitation regime ( @xmath20 ) .
goal of this study is to fill the gap between the two analytically derived extremal cases of very weak and very strong excitation as presented by boehme and lips @xcite and to describe a general behavior of the nutation frequency reflected by the spin
dependent transition rates for arbitrary @xmath12 and @xmath21 and arbitrary excitation frequencies @xmath22 . for a straight forward interpretation of experimental results @xcite
, it is of particular interest to understand if the change of rabi frequency from @xmath23 to @xmath24 takes place continuously or abruptly . as a function of the pulse length @xmath25 and the applied microwave frequency @xmath26 for a spin pair with large larmor separation @xmath27 .
the parameters of the simulation were @xmath28ghz , @xmath29ghz for @xmath30 and microwave excitation frequencies of @xmath31 as well as a @xmath12-field of @xmath32mhz .
one can distinguish the two resonant peaks and recognize the nutation on the pulse length axis.,width=317 ]
the basis for the results presented in the following are the pair models for spin dependent recombination and transport as described and discussed in detail in refs .
@xcite and @xcite , respectively .
these models are based on the kaplan
solomon mott model @xcite under consideration of non - negligible triplet transition probabilities and spin relaxation . for the calculation of the data presented we strictly follow these models under the assumption of negligible spin spin interactions .
note that this constraint does not apply to all known experimental systems and will always have to be considered when the results presented in the following are applied to the interpretation of experimental data .
quantitatively , the models outlined can be represented by an ensemble of spin @xmath16 pairs described by the density operator @xmath1 as derived in ref .
we define @xmath33 to be the hamiltonian of an individual pair with @xmath34 representing the unperturbed hamiltonian in the presence of a constant magnetic field @xmath35 and @xmath36 is the perturbation with a circularly polarized microwave of angular frequency @xmath22 and strength @xmath12 .
note the absence of spin spin coupling in @xmath37 in contrast to eq .
5 of ref .
the dynamics of the ensemble of spin pairs can be described by a liouville equation @xmath38 ^ -$ ] in which , in contrast to eq . 1 of ref .
@xcite , all stochastic terms have been dropped since incoherent processes are considered to be negligible during the applied coherent esr pulse . only coherent pulses these are pulses that are shorter than the fastest incoherent processes are considered in the following .
the ensemble of spins as represented by the density operator @xmath39 can be expressed by a unitary , time
dependent 4@xmath404 matrix . due to the absence of coupling
, the hamiltonian @xmath37 will be diagonal in the product base with four eigenvalues @xmath41 and @xmath42 wherein @xmath43 and @xmath21 represent the sum and the difference of the larmor frequencies @xmath44 and @xmath45 , respectively .
when a solution for @xmath46 is found , the transients of the spin
dependent transition rate @xmath47\ ] ] can be calculated from the projection of the permutation symmetry operators on the ensemble state .
experimentally , a real time observation of @xmath48 on typical time scales in the lower ns - range is difficult to obtain with podmr and often impossible with pedmr due to the insufficient time resolution caused by the long dielectric relaxation times of semiconductors in particular at low temperatures .
hence , transient nutation experiments are typically conducted by means of decay transient measurements as a function of the applied pulse length @xcite .
dynamics during the coherent spin excitations are obtained from these pulse length dependence measurements by charge integration @xmath49 which reveals the permutation symmetry state at the end of the microwave pulse as explained in detail in ref . @xcite . under the given conditions , @xmath50 } = \frac{\rho_{44}(\tau)-\rho^s_{44}}{\mathrm{tr}[\rho^s]}\ ] ] whereas @xmath51 is the steady state value of the matrix element @xmath52 of the density matrix @xmath1 .
hence , it is @xmath53 which is the observable calculated and displayed in the following . note
that while @xmath49 represents a number of charge carriers for pedmr experiments , the integration of the photoluminescence decay transient in podmr reveals a number of photons .
nevertheless , in both cases , the observable shall be referred to as @xmath49 in the following and is always plotted in arbitrary units because of this ambiguity .
when incoherence is negligible , the hamiltonian @xmath54 can be diagonalized for any given parameter set and thus , the liouville equation in its time integrated form @xmath55 can be solved by a simple matrix multiplication in which @xmath56 $ ] is the same initial state as used in ref .
the solution @xmath57 is referred to as `` semi analytic '' since this is easy to perform by means of calculation of the eigenvalues and eigenvectors of @xmath58 with any given parameter set while attempting to solve it fully analytically for arbitrary variables leads to unreasonably lengthy expressions .
figure [ fi2 ] displays an example for the time domain result for @xmath49 as a function of the larmor frequency for a spin pair ensemble with a larmor separation much larger than @xmath59 which is one of the extremal cases discussed in ref .
@xcite ( v.a.2 . ) .
one can recognize the two well distinguishable peaks at @xmath60ghz and @xmath61ghz on the larmor frequency
scale determined by the choice of the respective @xmath62-values of the spin
pair partners as well as the undamped rabi
nutation on the scale of the pulse length @xmath25 . outside of the resonances , the signal intensity drops while the nutation frequency increases .
the latter is well known and understood from the expression of the rabi frequency as introduced in eq .
28 of ref .
it is due to the spin nutation about the residual @xmath11-field @xmath63 in the rotating frame which increases with the distance of the excitation frequency from the resonance frequency .
figure [ fi2 ] shows that the relevant information contained in the calculated transients @xmath49 are the frequencies as well as the amplitudes of the nutation components .
thus , from the solutions of @xmath49 the absolute fourier transform ( ft ) was calculated in order to analyze the various nutation frequencies contained therein .
in contrast to the short 1@xmath64s transient shown in fig .
[ fi2 ] , the time scale for the time domain simulation and , therefore , the fourier integration was chosen to be 5@xmath64s long . for the parameters used throughout this study
this corresponds to at least 50 rabi oscillation periods .
figure [ fi3](a ) displays @xmath65 = ft@xmath66 obtained by semi
analytical calculations for spin pairs with four different larmor separations ( @xmath67 = 1mhz , 5mhz , 20mhz , 40mhz ) for microwave excitation frequencies of @xmath68 with a given @xmath12-field of @xmath32mhz .
while the plots ( a.i ) and ( a.iv ) fulfill the extremal cases of small and large larmor separation , respectively , the plots ( a.ii ) and ( a.iii ) describe two intermediate cases with @xmath69 .
note that the scaling of the color code was normalized to the maximum for each graph in order to achieve sufficient contrast . in order to be able to compare the four different cases more easily , the rabi components ft@xmath66 at the resonance frequency of the two spin partners
are plotted in the respective graphs ( red curves ) as conventional two - dimensional plots . here , the chosen scaling is equal for all graphs .
the four cases displayed in fig .
[ fi3](a ) confirm the hyperbolic increase of the rabi frequency @xmath70 as the microwave frequency is shifted out of resonance .
one can deduce from the two dimensional inset plots of fig .
[ fi3](a.i ) and ( a.iv ) , that the on resonance cases show only one frequency component for the two extremal cases , namely @xmath71 for large larmor separation and @xmath72 for small larmor separation .
this confirms the analytical results of ref .
@xcite . outside of the resonances ( @xmath73 )
, the oscillation splits into two components for the extremal cases and in the general , intermediate cases , there are up to four different nutation frequencies . in order to illustrate the transition from a single @xmath74 to a single @xmath72 frequency component with decreasing larmor separation ( or equivalently spoken , for an increasing microwave field @xmath12 ) , a plot of @xmath65 = ft@xmath66 versus the ratio @xmath75 on a logarithmic scale around @xmath76 is shown in fig .
[ fi4 ] for the one spin on resonant cases @xmath77 ( a ) and the average frequency case @xmath78 ( b ) .
again , for the extremal cases of @xmath79 and @xmath80 , the plots confirm the known results for small and large larmor separation described above . in fig .
[ fi4](a ) , one can see for @xmath79 ( @xmath81 < 0 $ ] ) that there is only one component with @xmath72 . as @xmath75 increases and approaches 1 ( @xmath82 $ ] approaches 0 )
, this component gradually becomes weaker and its frequency increases .
this behavior can be understood by the fact that in fig .
[ fi4](a ) , @xmath22 is always equal to one of the pair partner resonances .
when @xmath21 becomes larger , @xmath83 and thus , an increase of the observed nutation frequency takes place .
when @xmath84 ( @xmath85\approx 0 $ ] ) , two new nutation components become visible : ( i ) a low frequency component which generally is hard to separate from any given offset in the function @xmath49 and ( ii ) one component with @xmath71 .
the magnitude of the latter rises from very small values for @xmath79 and increases asymptotically to a maximum value for @xmath86 whereas the two other components vanish . a complementary view on these changes is given by fig .
[ fi4](b ) where the rabi components for an excitation frequency @xmath87 are plotted versus @xmath88 . for @xmath79 , plot ( a ) and ( b ) agree since they represent the same physical situation .
when @xmath75 increases and @xmath89 , no low frequency component becomes visible .
the @xmath71 component , which also becomes visible will increase proportionally to @xmath21 since it is off - resonant to the applied microwave frequency @xmath22 .
as function of @xmath90 for an excitation frequency ( a ) @xmath77 on resonance with one of the pair partners and ( b ) @xmath87 on resonance with the average @xmath91 of the pair partners larmor frequencies.,width=317 ]
the simulation of spin rabi oscillation as observed by pedmr / podmr reveals that the doubling of the nutation frequency @xmath71 to @xmath72 with decreasing larmor separation ( @xmath75 ) is abrupt which means that there is no continuous increase of the oscillation frequency . instead , only the magnitudes of the various components change in the intermediate larmor separation regime about @xmath92 . here , four nutation components become visible which can all become significant at the same time .
mathematically , it is obvious that the pairs consisting of two @xmath16 particles with four eigenstates will reveal four eigenfrequencies for any given general set of parameters .
we interpret this behavior for the general case as the interplay of the two one spin systems and the one two
spin system by which the spin pairs are described in the two extremal cases discussed above .
the results of the analytical derivation given in ref .
@xcite showed that the most significant qualitative change that takes place when the larmor separation changes from infinity to zero , is that a transition from a one spin to a two spin system occurs .
for the one spin system , the spin in resonance ( note that here , only one spin can be in resonance ) fully determines the oscillation of the pairs permutation symmetry whereas for the two spin system , the permutation symmetry is determined by the relative movements and spin orientations within the pair ensemble . for large larmor separation , at least one spin will remain fixed in its initial state , typically an eigenstate with polarization parallel to the @xmath11-axis . as the excitation frequency
is changed , the system becomes off - resonant , and two one spin contributions @xmath93 appear where the frequency of one increases whereas the frequency of the other decreases at the same time .
the pair still behaves like two individual one spin systems . for small larmor separation ,
there are always two spins in motion as long as the system is on resonance ( there is then only one resonance line observed ) .
hence , since the relative spin motion of the two spins within the pair will now determine the oscillation of the electronic rate transition , the beat frequencies @xmath94 of the two one spin nutation frequencies can be expected . here , @xmath95 and @xmath96 stand for parallel and antiparallel orientations , respectively .
figure [ fi3](b ) displays four plots of the nutation frequencies obtained with the simple terms given in eqs .
[ onespin ] and [ twospin ] for the same parameter sets used for the simulation results displayed in fig . [ fi3](a ) .
while this purely phenomenological description of the nutation frequencies can not account for the intensities of the nutation components this is the reason why there is no grey scale gradient in fig .
[ fi3](b ) it nevertheless shows , that there is an excellent agreement of eqs . [ onespin ] and [ twospin ] with the semi analytically calculated frequency patterns displayed in fig . [
fi3](b ) .
while simple quantitative predictions can be made for the nutation frequencies , there are no straight forward formulas for the prediction of the nutation amplitudes . as shown in fig .
[ fi3](a ) , the intensities of the different nutation components can have quite complex microwave frequency ( @xmath22 ) dependencies .
well separated lines ( case of large larmor separation ) exhibit lorentzian line shapes determined by the @xmath12 field due to the power broadening as described in ref . @xcite . for intermediate cases ,
the @xmath22dependence becomes much more complex and a given nutation component can exhibit a maximum at its corresponding resonance but also a local minimum at the resonance of a non corresponding transition .
an example for this behavior is the data of fig .
[ fi3](a.iii ) . at a frequency of @xmath97
one can recognize both , a maximum of the nutation component at @xmath71 but also a minimum of the respective other nutation component at @xmath98 .
qualitatively , this behavior can be interpreted by consideration of a four level system .
any of the four levels can undergo first order transitions into two different states . for excitations which are out of resonance with both transitions ,
the transition probability is small but may not be negligible in the vicinity of the resonances .
however , when one transition is induced resonantly , the transition into the non resonantly excited state is quenched at the same time and the intensity of its corresponding nutation component is quenched .
note that in spite of this qualitative interpretation , quantitative predictions have to be made by means of the simulation methods described above .
in summary , the response of charge carrier transport and recombination rates through localized electronic states in semiconductors to a coherent manipulation by magnetic resonance were simulated as they would be expected in pedmr / podmr experiments .
the transient response was calculated with the spin excitation present for different excitation fields and frequencies as well as different larmor separations within the pairs .
it was assumed that exchange and dipolar interaction are weak and incoherence due to the electronic transitions or spin - relaxation is negligible .
the presented data were obtained by a semi - analytical simulation method .
the results show that four qualitatively distinguishable nutation processes influence the oscillation of the transition rates which reduce to one significant contribution in the cases of large and small larmor separations .
simple empirical equations for the calculation of these nutation frequencies have been obtained which match the simulated data excellently and a qualitative picture for the interpretation of the nutation intensities has been discussed .
the presence of the four nutation processes implies that changing the larmor separation or the applied excitation field leads to an abrupt and not continuous change of the observed nutation frequencies .
10 b. c. cavenett , advance in physics * 30 * , 475 ( 1981 ) .
, m. s. brandt , and m.w .
bayerl , j. non cryst .
solids * 266 - 269 * , 1 ( 2000 ) . ,
i. martin , e. yablonovitch , and h. w. jiang , nature ( london ) * 430 * , 435 ( 2004 ) .
, t. gaebel , i. popa , a. gruber , and j. wrachtrup , phys .
lett . * 92 * , 076401 ( 2004 ) . , i. solomon , and n. f. mott , j. phys .
( paris ) lettres * 39 * , l51 ( 1978 ) .
b * 6 * , 436 ( 1972 ) . , w. r. buchwald , e. h. poindexter , w. l. warren , and d. j. keeble , solid state electron . *
34 * , 835 ( 1991 ) . and k. lips , phys .
c * 1 * , 1255 ( 2004 ) . , c. boehme , and k. lips , j. appl .
phys . * 97 * , 056101 ( 2005 ) . , s. roth , and m. mehring , mol . phys . * 95 * , 967 ( 1998 ) . and w. fuhs , j. appl . phys .
* 74 * , 3993 ( 1993 ) . and w. dietz , solid state commun . * 35 * , 505 ( 1980 ) . and k. lips , phys .
lett . * 91 * , 246603 ( 2003 ) . and k. lips , phys . rev .
b * 68 * , 245105 ( 2003 ) .
d. j. gravesteijn and m. glasbeek , phys .
b * 19 * , 5549 ( 1979 ) .
, c. von borczyskowski , j. bernard , m. orrit , and r. brown , nature * 363 * , 244 ( 1993 ) . , j. a. j. m. disselhorst , m. c. j. m. donckers , e. j. j. groenen , j. schmidt , and w. e. moerner , nature * 363 * , 242 ( 1993 ) . and h. overhof , _ point defects in semiconductors and insulators _ ( springer , berlin , 2002 ) . , c. boehme , and t. ehara , journal of optoelectronics and advanced materials * 7 * , 13 ( 2004 ) . and
g. jeschke , _ principles of pulse electron paramagnetic resonance _ ( oxford university press , oxford , 2001 ) . and k. lips , in _ charge transport in disordered solids with applications in electronics _ , edited by s. baranovski ( john wiley and sons , ltd . ,
chichester , england , 2006 ) , chap . 5 , pp .
c. boehme and k. lips , physica b * 376377 * , 930 ( 2006 ) . | semi analytical predictions for the transients of spin
dependent transport and recombination rates through localized states in semiconductors during coherent electron spin excitation are made for the case of weakly spin coupled charge carrier ensembles .
the results show that the on resonant rabi frequency of electrically or optically detected spin
oscillation doubles abruptly as the strength of the resonant microwave field @xmath0 exceeds the larmor frequency separation within the pair of charge carrier states between which the transport or recombination transition takes place . for the case of a larmor frequency separation of the order of @xmath0 and arbitrary excitation frequencies , the charge carrier pairs exhibit four different nutation frequencies . from the calculations ,
a simple set of equations for the prediction of these frequencies is derived . |
the heavy - light mesons listed in the review of particle physics can be classified into the spin doublets in the heavy quark limit , now the @xmath8 @xmath9 doublets @xmath10 , @xmath11 , @xmath12 , @xmath13 and the @xmath14 @xmath15 doublets @xmath16 , @xmath17 , @xmath18 , @xmath19 are complete @xcite .
the doublet @xmath17 are well - established experimentally , while the quantum numbers of the @xmath20 are not as well established , the width and decay modes are consistent with the @xmath21 assignment @xcite . in 2007 ,
the d0 collaboration firstly observed the @xmath22 and @xmath23 @xcite , later the cdf collaboration confirmed them , and obtained the width @xmath24 @xcite . also in 2007 , the cdf collaboration observed the @xmath25 and @xmath26 @xcite .
the d0 collaboration confirmed the @xmath26 @xcite . in 2012 ,
the lhcb collaboration updated the masses @xmath27 and @xmath28 , and measured the width @xmath29 @xcite .
recently , the cdf collaboration measured the masses and widths of the @xmath30 , @xmath31 , @xmath25 , @xmath32 , and observed a new excited state @xmath33 @xcite .
the @xmath14 @xmath15 doublets have been drawn little attention compared to the @xmath8 @xmath9 and @xmath14 @xmath34 heavy - light mesons @xcite .
we can study the masses , decay constants and strong decays of the @xmath14 @xmath15 doublets based on the qcd sum rules to obtain fruitful information about their internal structures and examine the heavy quark symmetry . the p - wave , d - wave and radial excited heavy - light mesons will be studied in details in the futures at the lhcb and kek - b .
experimentally , the strong decays of the @xmath14 @xmath35 doublets take place through relative d - wave , the corresponding widths are proportional to @xmath36 , with the angular momentum @xmath37 transferred in the decays . in these decays , the momentum @xmath38 is small , the decays are kinematically suppressed .
the strong decays @xmath39 , @xmath40 @xcite , @xmath41 @xcite , @xmath42 @xcite , @xmath43 @xcite , @xmath44 , @xmath45 , @xmath46 , @xmath47 @xcite , @xmath48 , @xmath49 @xcite have been observed .
the qcd sum rules ( qcdsr ) is a powerful nonperturbative theoretical tool in studying the ground state hadrons , and has given many successful descriptions of the masses , decay constants , hadronic form - factors , hadronic coupling constants , etc @xcite .
the hadronic coupling constants in the @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 vertices have been studied with the three - point qcdsr @xcite , while the hadronic coupling constants in the @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath55 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 , @xmath61 , @xmath88 , @xmath89 , @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 vertices have been studied with the light - cone qcdsr @xcite .
the detailed knowledge of the hadronic coupling constants is of great importance in understanding the effects of heavy quarkonium absorptions in hadronic matter .
furthermore , the hadronic coupling constants play an important role in understanding final - state interactions in the heavy quarkonium ( or meson ) decays and in other phenomenological analysis .
some hadronic coupling constants , such as @xmath0 , @xmath1 , @xmath2 , @xmath3 , can be directly extracted from the experimental data as the corresponding strong decays are kinematically allowed , we can confront the theoretical predications to the experimental data in the futures . in ref.@xcite , k. azizi et al study the masses and decay constants of the tensor mesons @xmath104 and @xmath105 with the qcdsr by only taking into account the perturbative terms and the mixed condensates in the operator product expansion . in ref.@xcite
, we calculate the contributions of the vacuum condensates up to dimension-6 in the operator product expansion , study the masses and decay constants of the heavy tensor mesons @xmath104 , @xmath105 , @xmath106 , @xmath26 with the qcdsr .
the predicted masses of the @xmath104 , @xmath105 , @xmath106 , @xmath26 are in excellent agreement with the experimental data , while the ratios of the decay constants @xmath107 , where the exp denotes the experimental value @xcite . in ref.@xcite , k. azizi
et al calculate the hadronic coupling constants @xmath108 and @xmath109 with the three - point qcdsr by choosing the tensor structure @xmath110 , then study the strong decays @xmath111 and @xmath112 , the decay widths are too small to account for the experimental data , if the widths of the tensor mesons are saturated approximately by the two - body strong decays . in the article
, we take the decay constants of the heavy tensor mesons as input parameters @xcite , analyze all the tensor structures to study the vertices @xmath113 , @xmath114 , @xmath115 , @xmath116 with the three - point qcdsr so as to choose the pertinent tensor structures ( in this article , we choose the tensor structures @xmath117 and @xmath118 , which differ from the tensor structure @xmath110 chosen in ref.@xcite . ) , then obtain the corresponding hadronic coupling constants , and study the two - body strong decays @xmath4 , @xmath5 , @xmath6 , @xmath7 and try to smear the large discrepancy between the theoretical calculations and the experimental data @xcite .
the article is arranged as follows : we derive the qcdsr for the hadronic coupling constants in the vertices @xmath113 , @xmath114 , @xmath115 , @xmath116 in sect.2 ; in sect.3 , we present the numerical results and calculate the two body strong decays ; and sect.4 is reserved for our conclusions .
in the following , we write down the three - point correlation functions @xmath119 in the qcdsr , @xmath120 where @xmath121 and @xmath122 , the pseudoscalar currents @xmath123 ( @xmath124 ) interpolate the heavy ( light ) pseudoscalar mesons @xmath125 and @xmath126 ( @xmath127 and @xmath128 ) , respectively , the tensor currents @xmath129 interpolate the heavy tensor mesons @xmath104 , @xmath105 , @xmath106 and @xmath26 , respectively .
we can insert a complete set of intermediate hadronic states with the same quantum numbers as the current operators @xmath130 , @xmath123 and @xmath131 into the correlation functions @xmath119 to obtain the hadronic representation @xcite . after isolating the ground state contributions from the heavy tensor mesons @xmath132 , heavy pseudoscalar mesons @xmath133 and light pseudoscalar mesons @xmath134 , we get the following result , @xmath135p_{\mu}p_{\nu}\right\}+\cdots\ , , \nonumber \\ & = & \pi_1(p^2,p^{\prime2})g_{\mu\nu}+\pi_2(p^2,p^{\prime2})p^{\prime}_{\mu}p^{\prime}_{\nu}+\pi_3(p^2,p^{\prime2})\left(p_{\mu}p^{\prime}_{\nu}+p^{\prime}_{\mu}p_{\nu } \right)+\pi_4(p^2,p^{\prime2})p_{\mu}p_{\nu}+\cdots \ , , \nonumber\\\end{aligned}\ ] ] where @xmath136 , the decay constants @xmath137 , @xmath138 , @xmath139 and the hadronic coupling constants @xmath140 are defined by @xmath141 the @xmath142 are the polarization vectors of the tensor mesons with the following properties , @xmath143 in general , we expect that we can choose either component @xmath144 ( with @xmath145 ) of the correlations @xmath119 to study the hadronic coupling constants @xmath140 . in calculations , we observe that the tensor structures @xmath117 and @xmath146 are the pertinent tensor structures . in ref.@xcite ,
k. azizi et al take the tensor currents @xmath147 , which couple both to the heavy tensor mesons and heavy scalar mesons , some contaminations are introduced .
now , we briefly outline the operator product expansion for the correlation functions @xmath119 in perturbative qcd .
we contract the quark fields in the correlation functions @xmath119 with wick theorem firstly , @xmath148 where @xmath149 @xmath150 @xmath151 , the @xmath152 is the gell - mann matrix , the @xmath153 , @xmath154 , @xmath155 are color indexes @xcite .
we usually choose the full light quark propagators in the coordinate space . in the present case ,
the quark condensates and mixed condensates have no contributions , so we take a simple replacement @xmath156 to obtain the full @xmath157 quark propagators . in the leading order approximation , the gluon field @xmath158 in the covariant derivative
has no contributions as @xmath159 .
then we compute the integrals to obtain the qcd spectral density through dispersion relation .
the leading - order contributions @xmath160 can be written as @xmath161\gamma_5 \left [ \!\not\!{k}+\!\not\!{p } -\!\not\!{p^{\prime}}+ m_{q^\prime}\right]\gamma_{\mu\nu}\left [ \!\not\!{k}-\!\not\!{p^{\prime } } + m_q\right]\right\}}{\left[k^2-m_q^2\right]\left[(k+p - p^{\prime})^2-m_{q^\prime}^2\right]\left[(k - p^{\prime})^2-m_q^2\right]}\ , , \nonumber\\ & = & \int ds du \frac{\rho_{\mu\nu}}{(s - p^2)(u - p^{\prime2 } ) } \ , , \end{aligned}\ ] ] where @xmath162 we put all the quark lines on mass - shell using the cutkosky s rules , see fig.1 , and obtain the leading - order spectral densities @xmath163 , @xmath164\delta\left[(k+p - p^{\prime})^2-m_{q^\prime}^2\right]\delta\left[(k - p^{\prime})^2-m_q^2\right]\nonumber\\ & & { \rm tr}\left\ { \gamma_5\left [ \!\not\!{k}+m_{q}\right]\gamma_5 \left [ \!\not\!{k}+\!\not\!{p } -\!\not\!{p^{\prime}}+ m_{q^\prime}\right]\gamma_{\mu\nu}\left [ \!\not\!{k}-\!\not\!{p^{\prime } } + m_q\right]\right\ } \ , , \\ &
= & \frac{g_{\mu\nu}}{4\pi^2\sqrt{\lambda(s , u , q^2)}}\left\ { m_q^3(m_{q^\prime}-m_q)-q^2m_q(m_q+m_{q^\prime})+m_q(sm_q - um_{q^\prime})\right .
\nonumber\\ & & \left.+ 6\left ( u - s+q^2 + 2m_qm_q-2m_{q^\prime}m_q\right)d_2(0,0,m_q)\right\ } \nonumber\\ & & + \frac{3p^\prime_{\mu}p^\prime_{\nu}}{2\pi^2\sqrt{\lambda(s , u , q^2)}}\left\ { u+q^2-m_q^2 + 2m_qm_q\right .
\nonumber\\ & & + ( s-2u-2q^2+m_q^2 - 4m_qm_q+2m_{q^\prime}m_q)b_1(0,0,m_q ) \nonumber\\ & & \left.+ \left ( u - s+q^2 + 2m_qm_q-2m_{q^\prime}m_q\right)b_2(0,0,m_q)\right\}+\cdots \ , , \end{aligned}\ ] ] where we have used the following formulae , @xmath165 \ , , \nonumber\\ \int d^4k\ , \delta^3 \
, k_{\mu } k_\nu&=&\frac{\pi}{2\sqrt{\lambda(s , u , q^2)}}\left[a_2(m_a , m_b , m_q)p_{\mu}p_\nu+b_2(m_a , m_b , m_q)p^\prime_{\mu}p^\prime_{\nu } \right .
\nonumber\\ & & \left .
+ c_2(m_a , m_b , m_q)\left(p_{\mu}p^\prime_\nu+p^\prime_{\mu}p_\nu\right)+d_2(m_a , m_b , m_q)g_{\mu\nu } \right]\ , , \end{aligned}\ ] ] @xmath166\delta\left[(k+p - p^\prime)^2-m_b^2 \right]\delta\left[(k - p^\prime)^2-m_q^2 \right ] \ , , \nonumber\\ b_1(m_a , m_b , m_q)&= & \frac{1}{\lambda(s , u , q^2)}\left [ m_q^2(s - u+q^2)+u(u - s-2q^2)+q^2(q^2-s)\right.\nonumber\\ & & \left.-2sm_a^2+m_b^2(u+s - q^2)\right]\ , , \nonumber\\ b_2(m_a , m_b , m_q)&=&\frac{1}{\lambda(s , u , q^2)}\left [ ( u - q^2-m_q^2)^2 + 2m_b^2(u - q^2-m_q^2)-4sm_a^2\right]\nonumber\\ & & + \frac{6s } { \lambda^2(s , u , q^2 ) } \left\ { q^2\left[m_q^4-(u+s - q^2)m_q^2+su \right]+m_a^2m_b^2(q^2-u - s)\right.\nonumber\\ & & + m_a^2\left [ s(s - u - q^2)+m_q^2(u - s - q^2)\right]\nonumber\\ & & \left.+m_b^2\left [ u(u - s - q^2)+m_q^2(s - u - q^2)\right]\right\}\ , , \nonumber\\ d_2(m_a , m_b , m_q)&= & \frac{1 } { 2\lambda(s , u , q^2 ) } \left\ { q^2\left[m_q^4-(u+s - q^2)m_q^2+su \right]+m_a^2m_b^2(q^2-u - s)\right.\nonumber\\ & & + m_a^2\left [ s(s - u - q^2)+m_q^2(u - s - q^2)\right]\nonumber\\ & & \left.+m_b^2\left [ u(u - s - q^2)+m_q^2(s - u - q^2)\right]\right\}\ , , \end{aligned}\ ] ] here we have neglected the terms @xmath167 and @xmath168 as they are irreverent in present calculations .
the gluon condensate contributions shown by the feynman diagrams in fig.2 are calculated accordingly .
we take quark - hadron duality below the continuum thresholds @xmath169 and @xmath170 respectively , and perform the double borel transform with respect to the variables @xmath171 and @xmath172 to obtain the qcdsr , @xmath173 \nonumber\\ & & + \frac{1}{\sqrt{\lambda(s , u , q^2)}}\langle\frac{\alpha_sgg}{\pi}\rangle\left[\frac{1}{9s}-\frac{s - u-3q^2}{12}\frac{\partial^2}{\partial m_a^2\partial m_b^2}d_2(m_a , m_b , m_q)\right.\nonumber\\ & & -\frac{s-3u - q^2}{12}\frac{\partial^2}{\partial m_a^2\partial m_q^2}d_2(m_a,0,m_q)+\frac{s+u+q^2}{12}\frac{\partial^2}{\partial m_b^2\partial m_q^2}d_2(0,m_b , m_q)\nonumber\\ & & \left.\left.-\frac{1}{3}\frac{\partial}{\partial m_a^2}d_2(m_a,0,m_q)-\frac{1}{2}\frac{\partial}{\partial m_b^2}d_2(0,m_b , m_q)-\frac{1}{2}\frac{\partial}{\partial m_q^2}d_2(0,0,m_q)\right]\right\}\ , , \end{aligned}\ ] ] @xmath174 \nonumber\\ & & + \frac{1}{\sqrt{\lambda(s , u , q^2)}}\langle\frac{\alpha_sgg}{\pi}\rangle\left [ -\frac{s - u-3q^2}{12}\frac{\partial^2}{\partial m_a^2\partial m_b^2}b_2(m_a , m_b , m_q)\right.\nonumber\\ & & -\frac{s-3u - q^2}{12}\frac{\partial^2}{\partial m_a^2\partial m_q^2}b_2(m_a,0,m_q)+\frac{s+u+q^2}{12}\frac{\partial^2}{\partial m_b^2\partial m_q^2}b_2(0,m_b , m_q)\nonumber\\ & & -\frac{1}{3}\frac{\partial}{\partial m_a^2}b_2(m_a,0,m_q)-\frac{1}{2}\frac{\partial}{\partial m_b^2}b_2(0,m_b , m_q)-\frac{1}{2}\frac{\partial}{\partial m_q^2}b_2(0,0,m_q ) \nonumber\\ & & \left.\left.+\frac{5}{6}\frac{\partial}{\partial m_a^2}b_1(m_a,0,m_q)+\frac{11}{12}\frac{\partial}{\partial m_b^2}b_1(0,m_b , m_q)+\frac{11}{12}\frac{\partial}{\partial m_q^2}b_1(0,0,m_q)\right]\right\ } \ , , \nonumber\\\end{aligned}\ ] ] where @xmath175 @xmath176 and @xmath177 , @xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath183 .
the hadronic input parameters are taken as @xmath184 , @xmath185 , @xmath186 , @xmath187 , @xmath188 , @xmath189 , @xmath190 , @xmath191 , @xmath192 , @xmath193 , @xmath194 , @xmath195 , @xmath196 , @xmath197 , @xmath198 from the particle data group @xcite .
the threshold parameters are taken as @xmath199 , @xmath200 , @xmath201 , @xmath202 , @xmath203 , @xmath204 from the qcdsr @xcite .
then the energy gaps @xmath205 , the contributions of the ground states are fully included .
the value of the gluon condensate @xmath206 is taken as the standard value @xmath207 @xcite .
the masses the @xmath208 and @xmath209 quarks are obtained through the gell - mann - oakes - renner relation @xmath210 , i.e. @xmath211 at the energy scale @xmath212 . in the article , we take the @xmath213 masses @xmath214 , @xmath215 and @xmath216 from the particle data group @xcite , and take into account the energy - scale dependence of the @xmath213 masses from the renormalization group equation , @xmath217^{\frac{4}{9 } } \ , , \nonumber\\ m_c(\mu)&=&m_c(m_c)\left[\frac{\alpha_{s}(\mu)}{\alpha_{s}(m_c)}\right]^{\frac{12}{25 } } \ , , \nonumber\\ m_b(\mu)&=&m_b(m_b)\left[\frac{\alpha_{s}(\mu)}{\alpha_{s}(m_b)}\right]^{\frac{12}{23 } } \ , , \nonumber\\ \alpha_s(\mu)&=&\frac{1}{b_0t}\left[1-\frac{b_1}{b_0 ^ 2}\frac{\log t}{t } + \frac{b_1 ^ 2(\log^2{t}-\log{t}-1)+b_0b_2}{b_0 ^ 4t^2}\right]\ , , \end{aligned}\ ] ] where @xmath218 , @xmath219 , @xmath220 , @xmath221 , @xmath222 , @xmath223 and @xmath224 for the flavors @xmath225 , @xmath226 and @xmath227 , respectively @xcite
. in ref.@xcite , we study the masses and decay constants of the heavy tensor mesons using the qcdsr , and obtain the values @xmath228 , @xmath229 , @xmath230 , @xmath231 , @xmath232 , @xmath233 , @xmath234 , @xmath235 . the predicted masses
@xmath236 , @xmath237 , @xmath238 and @xmath239 are in excellent agreement with the experimental data . in calculations ,
we take @xmath240 and @xmath241 for the charmed ( bottom ) tensor mesons @xcite , and evolve all the scale dependent quantities to the energy scales @xmath242 and @xmath243 respectively through the renormalization group equation .
the same energy scales and truncations in the operator product expansion lead to the values @xmath244 , @xmath245 , @xmath246 and @xmath247 .
if we take into account the perturbative corrections , the experimental values @xmath248 and @xmath249 can be reproduced @xcite . in this article , we take the values of the decay constants of the heavy - light mesons as @xmath250 , @xmath251 , @xmath252 , @xmath253 , @xmath254 and @xmath255 , and neglect the uncertainties so as to avoid doubling counting as the uncertainties originate mainly from the threshold parameters and heavy quark masses . from the qcdsr in eqs.(16 - 17 ) , we can see that there are no contributions come from the quark condensates and mixed condensates , and no terms of the orders @xmath256 , @xmath257 , @xmath258 , @xmath259 , @xmath180 , which are needed to stabilize the qcdsr so as to warrant a platform . in this article , we take the local limit @xmath260 , and obtain the local qcdsr .
the ground states , higher resonances and continuum states have the same weight @xmath261 , we use the threshold parameters ( or the cut - off ) @xmath169 and @xmath170 to avoid the contaminations of the higher resonances and continuum states , while the threshold parameters @xmath169 and @xmath170 are determined by the conventional qcdsr @xcite . at the qcd side , there are not terms of the orders @xmath256 , @xmath257 , @xmath258 , @xmath259 , which vanish in the limit @xmath260 , so the threshold parameters @xmath169 and @xmath170 survive in the local qcdsr .
now we obtain the hadronic coupling constants @xmath262 at the large space - like regions , for example , @xmath263 , then fit the hadronic coupling constants @xmath264 into the functions @xmath265 , where @xmath266 , the c , u , and l denote the central values , upper bound and lower bound , respectively , the numerical values are shown the table 1 . if the heavy quark symmetry and chiral symmetry work well , the physical values of the hadronic coupling constants should have the relations , @xmath267 from table 1 , we can see that the ratio , @xmath268 which is smaller than the expectation 1 . in calculations ,
we have used the @xmath269-quark mass @xmath270 at the energy scale @xmath271 , if we take larger value ( the value of the @xmath272 varies in a rather large range @xcite ) , say @xmath273 , the relations in eq.(21 ) can be satisfied .
so in this article , we prefer the values @xmath274 and @xmath275 from the qcdsr as they suffer from much less uncertainties induced by the light quark masses , and take the approximation @xmath276 and @xmath277 according to the heavy quark symmetry and chiral symmetry .
the perturbative qcd spectral densities associate with the tensor structure @xmath117 have dimension ( of mass ) 2 , while the perturbative qcd spectral densities associate with the tensor structure @xmath278 have dimension 0 , it is more reliable to take the perturbative qcd spectral densities associate with the tensor structure @xmath117 as they can embody the energy dependence efficiently .
the values of the hadronic coupling constants come from the qcdsr associate with the tensor @xmath117 are much larger than that of the tensor @xmath278 . in this article
, we prefer the values @xmath279 , @xmath280 associate with the tensor @xmath117 , as they can also lead to much larger decay widths and favor accounting for the experimental data . .
the parameters of the hadronic coupling constants @xmath264 , where the @xmath117 and @xmath278 denote the tensor structures of the qcdsr , the units of the @xmath264 , @xmath281 , @xmath282 and @xmath283 are @xmath284 , @xmath284 , @xmath285 and @xmath286 , respectively .
[ cols="^,^,^,^,^,^,^",options="header " , ] we can take the hadronic coupling constants @xmath287 as basic input parameters and study the following strong decays , @xmath288 which take place through relative d - wave . the decay widths can be written as @xmath289 where @xmath290 ( or @xmath291 ) for the final states @xmath292 , @xmath128 ( or @xmath293 ) .
the numerical results are @xmath294 from the experimental data of the babar collaboration , @xmath295 @xmath296 we can obtain the average , @xmath297 which is consistent with the pdg s average @xmath298 @xcite .
we assume @xmath299 and saturate the total decay width @xmath300 with the two - body strong decays @xmath301 , @xmath302 , @xmath303 , @xmath304 , then obtain the theoretical value , @xmath305 which is much smaller than the experimental value , @xmath306 the strong decays @xmath307 are greatly suppressed in the phase - space , while the strong decays @xmath308 violate the isospin conservation and are also greatly suppressed .
we saturate the total decay width @xmath309 with the two - body strong decays @xmath310 , @xmath311 , and obtain the theoretical value , @xmath312 which is smaller than the experimental value , @xmath313 at the bottom sector , we assume @xmath314 according to the experimental value @xcite @xmath315 and neglect the kinematically suppressed decays @xmath316 and isospin violated decays @xmath317 , and saturate the total decay widths @xmath318 and @xmath319 with the two - body strong decays @xmath320 and @xmath321 , respectively . then we obtain the theoretical values , @xmath322 which are smaller than the experimental values , @xmath323 the perturbative @xmath324 corrections increase the correlation function ( or the product @xmath325 ) about @xmath326 in the light - cone qcd sum rules for the hadronic coupling constant @xmath327 @xcite . in the present case , we can assume the perturbative @xmath324 corrections also increase the correlation functions ( or the products @xmath328 ) about @xmath326 .
the perturbative @xmath324 corrections to the decay constants @xmath137 are negative @xcite , the net perturbative @xmath324 corrections to the @xmath329 are larger than @xmath326 . if half of those perturbative @xmath324 corrections are compensated by the perturbative @xmath324 corrections to the decay constants @xmath138 , the hadronic coupling constants @xmath140 are increased by about @xmath330 , then taking into account the perturbative @xmath324 corrections lead to the following replacements , @xmath331 then the theoretical values @xmath332 , @xmath333 and @xmath334 are compatible with the experimental data , while the theoretical value @xmath335 is still smaller than the experimental value .
in the article , we choose the pertinent tensor structures to calculate the hadronic coupling constants @xmath0 , @xmath1 , @xmath2 , @xmath3 with the three - point qcdsr , then study the two body strong decays @xmath4 , @xmath5 , @xmath6 , @xmath7 , the predicted total widths are compatible with the experimental data , while the predicted partial widths can be confronted with the experimental data from the besiii , lhcb , cdf , d0 and kek - b collaborations in the futures .
we can also take the hadronic coupling constants as basic input parameter in many phenomenological analysis .
this work is supported by national natural science foundation , grant numbers 11375063 , the fundamental research funds for the central universities , and natural science foundation of hebei province , grant number a2014502017 .
f. s. navarra , m. nielsen , m. e. bracco , m. chiapparini and c. l. schat , phys . lett .
* b489 * ( 2000 ) 319 ; m. e. bracco , m. chiapparini , a. lozea , f. s. navarra and m. nielsen , phys . lett .
* b521 * ( 2001 ) 1 ; f. s. navarra , m. nielsen and m. e. bracco , phys . rev .
* d65 * ( 2002 ) 037502 ; r. d. matheus , f.s .
navarra , m. nielsen and r. rodrigues da silva , phys . lett .
* b541 * ( 2002 ) 265 ; r. rodrigues da silva , r. d. matheus , f. s. navarra and m. nielsen , braz . j. phys .
* 34 * ( 2004 ) 236 ; m. e. bracco , m. chiapparini , f. s. navarra and m. nielsen , phys . lett . * b605 * ( 2005 ) 326 ; m. e. bracco , a. cerqueira , m. chiapparini , a. lozea and m. nielsen , phys . lett . * b641 * ( 2006 ) 286 ; m. e. bracco , m. chiapparini , f. s. navarra and m. nielsen , phys . lett . * b659 * ( 2008 ) 559 ; m. e. bracco and m. nielsen , phys . rev .
* d82 * ( 2010 ) 034012 ; b. o. rodrigues , m. e. bracco , m. nielsen and f. s. navarra , nucl . phys . * a852 * ( 2011 ) 127 ; k. azizi and h. sundu , j. phys .
* g38 * ( 2011 ) 045005 ; h. sundu , j.y .
sungu , s. sahin , n. yinelek and k. azizi , phys . rev .
* d83 * ( 2011 ) 114009 ; a. cerqueira , jr , b. o. rodrigues and m. e. bracco , nucl . phys .
* a874 * ( 2012 ) 130 ; c. y. cui , y. l. liu and m. q. huang , phys . lett . * b707 * ( 2012 ) 129 ; c. y. cui , y. l. liu and m. q. huang , phys . lett .
* b711 * ( 2012 ) 317 ; z. g. wang , phys . rev . *
d89 * ( 2014 ) 034017 .
p. colangelo , f. de fazio , g. nardulli , n. di bartolomeo and r. gatto , phys . rev .
* d52 * ( 1995 ) 6422 ; t. m. aliev , n. k. pak and m. savci , phys .
b390 * ( 1997 ) 335 ; p. colangelo and f. de fazio , eur .
j. * c4 * ( 1998 ) 503 ; y. b. dai and s. l. zhu , phys . rev .
* d58 * ( 1998 ) 074009 ; s. l. zhu and y. b. dai , phys . rev .
* d58 * ( 1998 ) 094033 ; a. khodjamirian , r. ruckl , s. weinzierl and o. i. yakovlev , phys . lett . *
b457 * ( 1999 ) 245 ; z. h. li , t. huang , j. z. sun and z. h. dai , phys . rev .
* d65 * ( 2002 ) 076005 ; h. c. kim and s. h. lee , eur .
j. * c22 * ( 2002 ) 707 ; d. becirevic , j. charles , a. leyaouanc , l. oliver , o. pene and j. c. raynal , jhep * 0301 * ( 2003 ) 009 ; z. g. wang and s. l. wan , phys .
* d73 * ( 2006 ) 094020 ; z. g. wang and s. l. wan , phys
. rev . * d74 * ( 2006 ) 014017 ; z. g. wang , eur .
j. * c52 * ( 2007 ) 553 ; z. g. wang , nucl . phys .
* a796 * ( 2007 ) 61 ; z. g. wang , j. phys . * g34 * ( 2007 ) 753 ; z. g. wang , phys .
* d77 * ( 2008 ) 054024 ; z. g. wang and z. b. wang , chin .
* 25 * ( 2008 ) 444 ; z. h. li , w. liu and h. y. liu , phys . lett .
* b659 * ( 2008 ) 598 . | in the article , we calculate the hadronic coupling constants @xmath0 , @xmath1 , @xmath2 , @xmath3 with the three - point qcd sum rules , then study the two body strong decays @xmath4 , @xmath5 , @xmath6 , @xmath7 , and make predictions to be confronted with the experimental data in the future . + zhi - gang wang + department of physics , north china electric power university , baoding 071003 , p. r. china pacs number : 13.20.fc , 13.20.he key words : hadronic coupling constants , tensor mesons , qcd sum rules |
in this paper , we consider the following equilibrium problem ( @xmath1 ) in the sense of blum and otteli @xcite , which consists in finding a point @xmath2 such that @xmath3 where @xmath4 is a nonempty , closed and convex subset of a real banach space @xmath5 and @xmath6 is an equilibrium bifunction , i.e. , @xmath7 for all @xmath8 . the solution set of ( @xmath1 ) is denoted by @xmath9 .
the equilibrium problem which also known under the name of ky fan inequality @xcite covers , as special cases , many well - known problems , such as the optimization problem , the variational inequality problem and nonlinear complementarity problem , the saddle point problem , the generalized nash equilibrium problem in game theory , the fixed point problem and others ; ( see @xcite ) .
also numerous problems in physic and economic reduce to find a solution of an equilibrium problem .
many methods have been proposed to solve the equilibrium problems see for instance @xcite . in 1980,cohen
@xcite introduced a useful tool for solving optimization problem which is known as auxiliary problem principle and extended it to variational inequality @xcite .
in auxiliary problem principle a sequence @xmath10 is generated as follows : @xmath11 is a unique solution of the following strongly convex problem @xmath12 where @xmath13 .
recently , mastroeni @xcite extended the auxiliary problem principle to equilibrium problems under the assumptions that the equilibrium function @xmath14 is strongly monotone on @xmath15 and that @xmath14 satisfies the following lipschitz- type condition : @xmath16 for all @xmath17 where @xmath18 . to avoid the monotonicity of @xmath14 , motivated by antipin @xcite , tran et al .
@xcite have used an extrapolation step in each iteration after solving ( [ equ.1 ] ) and suppose that @xmath14 is pseudomonotone on @xmath15 which is weaker than monotonicity assumption .
they assumed @xmath19 was the unique solution of ( [ equ.1 ] ) and the unique solution of the following strongly convex problem @xmath20 is denoted by @xmath21 .
in special case , when the problem @xmath22 is a variational inequality problem , this method reduces to the classical extragradient method which has been introduced by korpelevich @xcite .
the extragradient method is well known because of its efficiency in numerical tests . in the recent years
, many authors obtained extragradient algorithms for solving ( @xmath1 ) in hilbert spaces where convergence of the proposed algorithms was required @xmath14 to satisfy a certain lipschitz - type condition @xcite .
lipschitz - type condition depends on two positive parameters @xmath23 and @xmath24 which in some cases , they are unknown or difficult to approximate . in other to avoid this requirement , authors used the linesearch technique in a hilbert space to obtain convergent algorithms for solving equilibrium problem @xcite . in this paper , we consider the following auxiliary equilibrium problem ( @xmath25 ) for finding @xmath2 such that @xmath26 for all @xmath27 , where @xmath28 is a regularization parameter and @xmath29 be a nonnegative differentiable convex bifunction on @xmath4 with respect to the second argument @xmath30 , for each fixed @xmath8 , such that 1 . @xmath31 for all @xmath8 , 2 .
@xmath32 for all @xmath8 . where @xmath33 denotes the gradient of the function @xmath34 at @xmath35 . in the recent years , many authors studied the problem of finding a common element of the set of fixed points of a nonlinear mapping and the set of solutions of an equilibrium problem in the framework of hilbert spaces and banach spaces , see for instance @xcite . in all of these methods ,
authors have used metric projection in hilbert spaces and generalized metric projection in banach spaces . in this paper , motivated d. q. tran et al .
@xcite and p. t. vuong et al .
@xcite , we introduce new extragradient and linesearch algorithms for finding a common element of the set of solutions of an equilibrium problem and the set of fixed points of a relatively nonexpansive mapping in banach spaces , by using sunny generalized nonexpansive retraction . using this method
, we prove strong convergence theorems under suitable conditions .
we denote by @xmath36 the normalized duality mapping from @xmath5 to @xmath37 defined by @xmath38 also , denote the strong convergence and the weak convergence of a sequence @xmath39 to @xmath35 in @xmath5 by @xmath40 and @xmath41 , respectively .
+ let @xmath42 be the unite sphere centered at the origin of @xmath5 .
a banach space @xmath5 is strictly convex if @xmath43 , whenever @xmath44 and @xmath45 .
modulus of convexity of @xmath5 is defined by @xmath46 for all @xmath47 $ ] .
@xmath5 is said to be uniformly convex if @xmath48 and @xmath49 for all @xmath50 .
let @xmath51 be a fixed real number with @xmath52 .
a banach space @xmath5 is said to be @xmath51-uniformly convex @xcite if there exists a constant @xmath53 such that @xmath54 for all @xmath47 $ ] .
the banach space @xmath5 is called smooth if the limit @xmath55 exists for all @xmath56 .
it is also said to be uniformly smooth if the limit @xmath57 is attained uniformly for all @xmath56 .
every uniformly smooth banach space @xmath5 is smooth . if a banach space @xmath5 uniformly convex , then @xmath5 is reflexive and strictly convex @xcite .
many properties of the normalized duality mapping @xmath36 have been given in @xcite .
+ we give some of those in the following : 1 . for every @xmath58
, @xmath59 is nonempty closed convex and bounded subset of @xmath60 .
2 . if @xmath5 is smooth or @xmath60 is strictly convex , then @xmath36 is single - valued .
if @xmath5 is strictly convex , then @xmath36 is one - one .
if @xmath5 is reflexive , then @xmath36 is onto .
if @xmath5 is strictly convex , then @xmath36 is strictly monotone , that is , @xmath61 for all @xmath62 such that @xmath45 . 6 .
if @xmath5 is smooth , strictly convex and reflexive and @xmath63 is the normalized duality mapping on @xmath60 , then @xmath64 , @xmath65 and @xmath66 , where @xmath67 and @xmath68 are the identity mapping on @xmath5 and @xmath60 , respectively .
if @xmath5 is uniformly convex and uniformly smooth , then @xmath36 is uniformly norm - to - norm continuous on bounded sets of @xmath5 and @xmath64 is also uniformly norm - to - norm continuous on bounded sets of @xmath60 , i.e. , for @xmath69 and @xmath70 , there is a @xmath71 such that @xmath72 @xmath73 let @xmath5 be a smooth banach space , we define the function @xmath74 by @xmath75 for all @xmath62 . observe that , in a hilbert space @xmath76 , @xmath77 for all @xmath78 + it is clear from definition of @xmath0 that for all @xmath79 , @xmath80 @xmath81 @xmath82 if @xmath5 additionally assumed to be strictly convex , then @xmath83 also , we define the function @xmath84 by @xmath85 , + for all @xmath58 and @xmath86
. that is , @xmath87 for all @xmath58 and @xmath88 . + it is well known that , if @xmath5 is a reflexive strictly convex and smooth banach space with @xmath89 as its dual , then @xmath90 for all @xmath58 and all @xmath91 @xcite .
let @xmath5 be a smooth banach space and @xmath4 be a nonempty subset of @xmath5 .
a mapping @xmath92 is called generalized nonexpansive @xcite if @xmath93 and @xmath94 for all @xmath8 and all @xmath95 + let @xmath4 be a closed convex subset of @xmath5 and @xmath96 be a mapping .
a point @xmath51 in @xmath4 is said to be an asymptotic fixed point of @xmath97 if @xmath4 contains a sequence @xmath39 which converges weakly to @xmath51 such that @xmath98 .
the set of asymptotic fixed points of @xmath97 will be denoted by @xmath99 .
a mapping @xmath96 is called relatively nonexpansive if @xmath100 and @xmath101 for all @xmath8 and @xmath102 .
the asymptotic behavior of relatively nonexpansive mappings was studied in @xcite .
@xmath97 is said to be relatively quasi - nonexpansive if @xmath103 and @xmath101 for all @xmath8 and all @xmath102 . the class of relatively quasi - nonexpansive mapping is broader than the class of relatively nonexpansive mappings which requires
+ it is well known that , if @xmath5 is a strictly convex and smooth banach space , @xmath4 is a nonempty closed convex subset of @xmath5 and @xmath96 is a relatively quasi - nonexpansive mapping , then @xmath104 is a closed convex subset of @xmath4 @xcite .
let @xmath105 be a nonempty subset of a banach space @xmath5 .
a mapping @xmath106 is said to be sunny @xcite if @xmath107 for all @xmath58 and all @xmath108 .
a mapping @xmath106 is said to be a retraction if @xmath109 for all @xmath110 .
@xmath111 is a sunny nonexpansive retraction from @xmath5 onto @xmath105 if @xmath111 is a retraction which is also sunny and nonexpansive .
a nonempty subset @xmath105 of a smooth banach space @xmath5 is said to be a generalized nonexpansive retract ( resp .
sunny generalized nonexpansive retract ) of @xmath5 if there exists a generalized nonexpansive retraction ( resp .
sunny generalized nonexpansive retraction ) @xmath111 from @xmath5 onto @xmath105 .
+ if @xmath5 is a smooth , strictly convex and reflexive banach space , @xmath112 be a nonempty closed convex subset of @xmath89 and @xmath113 be the generalized metric projection of @xmath89 onto @xmath112 . then the @xmath114 is a sunny generalized nonexpansive retraction of @xmath5 onto @xmath115 @xcite . if @xmath5 is a hilbert space
. then @xmath116 .
we need the following lemmas for the proof of our main results .
+ if @xmath4 is a convex subset of banach space @xmath5 , then we denote by @xmath117 the normal cone for @xmath4 at a point @xmath118 , that is @xmath119 suppose that @xmath5 is a banach space and let @xmath120 $ ] be a paper function .
for @xmath121 , we define the subdifferential of @xmath14 at @xmath122 as the subset of @xmath89 given by @xmath123 if @xmath124
, then we say @xmath14 is subdifferentiable at @xmath122 .
[ lem7 ] let @xmath4 be a nonempty convex subset of a banach space @xmath5 and @xmath125 be a convex and subdifferentiable function , then @xmath14 is minimized at @xmath58 if and only if @xmath126 [ lem8]@xcite let @xmath5 be a reflexive banach space . if @xmath127 and @xmath128 are nontrivial , convex and lower continuous functions and if @xmath129 , then @xmath130 [ lem20]@xcite
suppose that a convex function @xmath14 is continuouse on the interior of its domain .
then , for all @xmath131 , @xmath132 is non - empty and bounded .
[ lem13 ] let @xmath4 be a nonempty convex subset of a banach space @xmath5 and let @xmath133 be an equilibrium bifunction and convex respect to the second variable
. then the following statements are equivalent : 1 .
@xmath134 is a solution to @xmath9 , 2 .
@xmath134 is a solution to the problem @xmath135 using lemma [ lem7 ] , we get desired results .
equivalence between @xmath9 and ( @xmath25 ) is stated in the following lemma .
[ lem2 ] let @xmath4 nonempty , convex and closed subset of a reflexive banach space @xmath5 and @xmath133 be an equilibrium bifunction and let @xmath2 .
suppose that @xmath136 is convex and subdifferentiable on @xmath4 .
let @xmath137 be a differentiable convex function on @xmath4 with respect to the second argument @xmath30 such that 1 .
@xmath138 , 2 .
@xmath139 .
then @xmath2 is a solution to @xmath9 if and only if @xmath134 is a solution to ( @xmath25 ) .
it is clear from lemmas [ lem8 ] and [ lem13 ] .
@xcite let @xmath4 be a nonempty closed sunny generalized nonexpansive retract of a smooth and strictly convex banach space @xmath5 .
then the sunny generalized nonexpansive retraction from @xmath5 onto @xmath4 is uniquely determined .
@xcite[lem4 ] let @xmath4 be a nonempty closed subset of a smooth and strictly convex banach space @xmath5 such that there exists a sunny generalized nonexpansive retraction @xmath111 from @xmath5 onto @xmath4 and let @xmath140 .
then the following hold : 1 .
@xmath141 if and only if @xmath142 for all @xmath27 , 2 .
@xmath143 @xcite[eq38 ]
let @xmath5 be a smooth , strictly convex and reflexive banach space and @xmath4 be a nonempty closed sunny generalized nonexpansive retract of @xmath5 .
let @xmath111 be the sunny generalized nonexpansive retraction from @xmath5 onto @xmath4 and @xmath140
. then the following are equivalent : 1 .
@xmath144 , 2 .
@xmath145 @xcite[ibaraki ]
let @xmath5 be a smooth , strictly convex and reflexive banach space and let @xmath4 be a nonempty closed subset of @xmath5
. then the following are equivalent : 1 .
@xmath4 is a sunny generalized nonexpansive retract of @xmath5 , 2 .
@xmath4 is a generalized retract of @xmath5 , 3 .
@xmath146 is closed and convex .
@xcite[eq6 ] let @xmath5 be a @xmath147-uniformly convex and smooth banach space .
then , for all @xmath62 , we have @xmath148 where @xmath149 is the @xmath147-uniformly convex constant of @xmath5 .
[ lem16]@xcite suppose @xmath150 is a real number , then the following are equivalent 1 .
@xmath5 is a @xmath51-uniformly convex banach space , 2 .
there exits @xmath151 , such that for each @xmath152 and @xmath153 , we have @xmath154 where @xmath155 .
[ kamimra]@xcite let @xmath5 be a uniformly convex banach space and let @xmath156 .
then there exists a strictly increasing , continuous and convex function @xmath157\rightarrow [ 0 , \infty)$ ] , @xmath158 such that @xmath159 for all @xmath160 .
@xcite[lem15 ] let @xmath5 be a uniformly convex banach space .
then there exists a continuous strictly increasing convex function @xmath157\rightarrow [ 0 , \infty)$ ] , @xmath158 such that @xmath161 for all @xmath162 and all @xmath163 $ ] with @xmath164 .
@xcite[lem5 ] let @xmath5 be a uniformly convex and smooth banach space and let @xmath39 and @xmath165 be two sequences of @xmath5 . if @xmath166 and either @xmath39 or @xmath165 is bounded , then @xmath167 [ lem10]@xcite
let @xmath168 and @xmath169 be two positive and bounded sequences in @xmath170 , then @xmath171
in this section , we present an algorithm for finding a solution of the ( @xmath1 ) which is also the common element of the set of fixed points of a relatively nonexpansive mapping .
+ here , we assume that bifunction @xmath133 satisfies in following conditions which @xmath4 is nonempty , convex and closed subset of uniformly convex and uniformly smooth banach space @xmath5 , 1 .
@xmath7 for all @xmath172 2 .
@xmath14 is pseudomonotone on @xmath4 , i.e. , @xmath173 for all @xmath174 3 .
@xmath14 is jointly weakly continuous on @xmath15 , i.e. , if @xmath175 and @xmath176 and @xmath177 are two sequences in @xmath4 converging weakly to @xmath35 and @xmath30 , respectively , then @xmath178 4 .
@xmath179 is convex , lower semicontinuous and subdifferentiable on @xmath4 for every @xmath172 5 .
@xmath14 satisfies @xmath0-lipschitz - type condition : @xmath180 such that for every @xmath181 @xmath182 it is easy to see that if @xmath14 satisfies the properties @xmath183 , then the set @xmath9 of solutions of an equilibrium problem is closed and convex .
indeed , when @xmath5 is a hilbert space , @xmath0-lipschitz - type condition reduces to lipschitz - type condition ( [ equ.2 ] ) . + throughout the paper @xmath184 is a relatively nonexpansive self - mapping of @xmath4 .
* algorithm @xmath185 * step 0 .
: : suppose that @xmath186 $ ] for some @xmath187 , @xmath188 $ ] for some @xmath189 and @xmath190 $ ] such that @xmath191 $ ] , where @xmath192 step 1 .
: : let @xmath193 .
: : compute @xmath194 and @xmath195 , such that @xmath196 @xmath197 step 3 .
: : compute @xmath198 . + if @xmath199 and @xmath200 , then @xmath201 and go to step 4 .
: : compute @xmath202 , where @xmath203 is the sunny generalized nonexpansive retraction from @xmath5 onto @xmath204 and @xmath205 @xmath206 step 5 .
: : set @xmath207 and go to step 2 . before proving the strong convergence of the iterates generated by algorithm 1
, we prove following lemmas .
[ lem1 ] for every @xmath208 and @xmath209 , we obtain 1 .
@xmath210 2 .
@xmath211 . by the condition ( @xmath212 ) for @xmath179 and from lemmas [ lem7 ] and [ lem8 ]
, we obtain @xmath213 if and only if @xmath214 this implies that @xmath215 and @xmath216 exist such that @xmath217 so , from definition of @xmath218 , we obtain @xmath219 for all @xmath27 .
set @xmath220 , we have @xmath221 so , by definition of the @xmath222 and equality ( [ equa18 ] ) , we get @xmath223 for all @xmath27 . put @xmath220 in inequality [ eq40 ] , we have @xmath224 since @xmath225 and @xmath14 is pseudomonotone on @xmath4 . replacing @xmath226 and @xmath227 by @xmath195 , @xmath194 and @xmath228 in inequality ( [ equa17 ] ) , respectively ,
we get @xmath229 in a similar way , since @xmath230 , we have @xmath231 for all @xmath27 , hence ( @xmath232 ) is proved .
let @xmath233 in above inequality , we obtain @xmath234 combining inequalities ( [ equa1 ] ) , ( [ equa2 ] ) and ( [ equa3 ] ) , we get @xmath235 from inequality ( [ equa19 ] ) and ( [ eq36 ] ) , we have @xmath236 hence , ( @xmath237 ) is proved . in a real hilbert space @xmath5 , lemma [ lem1 ]
is reduced to lemma @xmath238 in @xcite . in algorithm @xmath185
, we obtain the unique optimal solutions @xmath194 and @xmath228 .
let @xmath239 , then using lemma [ lem1](@xmath232 ) , we have @xmath240 @xmath241 set @xmath242 in inequality ( [ equa22 ] ) and @xmath243 in inequality ( [ equa23 ] ) . hence , we get @xmath244 since @xmath36 is monotone and one - one , we obtain @xmath245 in a similar way , also @xmath228 is unique . if @xmath5 is a real hilbert space , then algorithm @xmath185 is the same extragradient algorithm in @xcite provided that the sequence @xmath168 satisfies the conditions of step @xmath246 of algorithm @xmath185 .
[ lem3 ] for every @xmath247 and @xmath209 , we get @xmath248 from lemma [ lem1 ] , by the convexity of @xmath249 and by the definition of functions @xmath0 and @xmath184 , we have @xmath250 we examine the stopping condition in the following lemma .
[ lem6 ] let @xmath199 , then @xmath251 . if @xmath199 and @xmath200 , then @xmath201 .
suppose @xmath199 , then by the definition of @xmath194 , condition ( @xmath252 ) , property of @xmath0 ( [ equa21 ] ) and since @xmath253 , we have @xmath254 for all @xmath27 .
set @xmath255 , hence lemma [ lem2 ] implies that @xmath251 .
+ let @xmath199 and @xmath200 , we have that @xmath256 and since @xmath257 is one - one , we get @xmath258 since @xmath259 and @xmath260 , it follows that @xmath261 and since @xmath36 is one - one , we get @xmath262 .
so @xmath263 in a real hilbert space @xmath5 , lemma [ lem6 ] is the same proposition @xmath264 in @xcite with different proof , provided that the sequence @xmath168 satisfies the conditions of step @xmath246 of algorithm @xmath185 .
[ eq25 ] let @xmath4 be a nonempty closed convex subset of a uniformly convex , uniformly smooth banach space @xmath5 .
assume that @xmath265 is a bifunction which satisfies conditions @xmath266 and @xmath267 is a relatively nonexpansive mapping such that @xmath268 then sequences @xmath269 , @xmath270 , @xmath271 and @xmath272 generated by algorithm @xmath185 converge strongly to the some solution @xmath273 , where @xmath274 , and @xmath275 is sunny generalized nonexpansive retraction from @xmath5 onto @xmath276 .
at first , using induction we show that @xmath277 for all @xmath278 .
let @xmath279 , from lemma [ lem3 ] , we get @xmath280 for all @xmath278 . now , we show that @xmath281 for all @xmath282 .
it is clear that @xmath283 .
suppose that @xmath281 , i.e @xmath284 for all @xmath285 .
since @xmath202 , using lemma [ lem4 ] , we get @xmath286 for all @xmath287 .
this implies that @xmath288 .
therefore @xmath289 .
let @xmath290 .
since @xmath291 , using successively equality ( [ eq36 ] ) , it is easy to see that the @xmath292 is increasing and bounded from above by @xmath293 , so @xmath294 exists .
this yields that @xmath292 is bounded . from inequality ( [ eq14 ] ) , we know that @xmath176 is bounded .
it is clear that @xmath295 , so lemma [ lem5 ] implies that @xmath296 and therefore @xmath176 converges strongly to @xmath297 since @xmath36 is uniformly norm - to - norm continuous on bounded sets , from equality ( [ equa6 ] ) , we obtain that @xmath298 , and @xmath299 exists for all @xmath300 . since @xmath301 , we have @xmath302 and from lemma [ lem5 ] , we deduce that @xmath303 , thus @xmath304 which implies that @xmath305 converges strongly to @xmath306 . using norm - to - norm continuity of @xmath36 on bounded sets ,
we conclude that @xmath307 and therefore @xmath308 using lemma [ lem1 ] @xmath309 , we obtain @xmath310 . from inequality ( [ eq14 ] ) and the definition of @xmath184 , we derive that @xmath311 and @xmath312 are bounded .
let @xmath313 and@xmath314 .
so , by lemma [ lem15 ] , there exists a continuous , strictly increasing and convex function @xmath315\rightarrow\mathbb{r}$ ] with @xmath316 such that for all @xmath285 , we get @xmath317 and using the same argument , there exists a continuous , strictly increasing and convex function @xmath318\rightarrow\mathbb{r}$ ] with @xmath319 such that for all @xmath285 , we have @xmath320 which imply @xmath321 @xmath322 by letting @xmath323 in inequalities ( [ equa8 ] ) and ( [ equa11 ] ) , using lemma [ lem10 ] and equality ( [ equa9 ] ) , we obtain @xmath324 from the properties of @xmath325 and @xmath326 , we have @xmath327 so from ( [ eq18 ] ) , we obtain @xmath328 since @xmath257 is uniformly norm - to - norm continuous on bounded sets . by the same reason as in the proof of ( [ equa9 ] ) , we can conclude from ( [ equa33 ] ) and ( [ equa12 ] ) that @xmath329 for all @xmath330 . using lemma [ lem1 ] @xmath309
, we have @xmath331 for all @xmath330 . taking the limits as @xmath323 in inequality ( [ equa35 ] ) , using equality ( [ equa34 ] ) , we get @xmath332 since @xmath176 and @xmath311 are bounded , it follows from lemma [ lem5 ] that @xmath333 which imply @xmath177 and @xmath311 converges strongly to @xmath297 now , we prove that @xmath334 .
it follows from the definition of @xmath194 that @xmath335 for all @xmath27 . by letting @xmath323 in inequality ( [ equa36 ] )
, it follows from equality ( [ equa10 ] ) , conditions @xmath336 and @xmath337 and uniformly norm - to - norm continuity of @xmath36 on bounded sets that @xmath338 because of @xmath253 . hence , letting @xmath339 , lemma [ lem2 ] implies that @xmath334 .
now , since @xmath340 , from ( [ equa12 ] ) , we get @xmath341 .
so , using the definition of @xmath184 , we have @xmath342 setting @xmath343 in lemma [ lem4 ] , since @xmath202 and @xmath0 is continuous respect to the first argument , we obtain @xmath344 also , using lemma [ eq38 ] , we have @xmath345 for all @xmath346 because of @xmath347
. therefore @xmath348 and consequently the sequences @xmath269 , @xmath270 , @xmath271 and @xmath272 converge strongly to @xmath349 .
if @xmath5 is a real hilbert space , then theorem [ eq25 ] is the same theorem @xmath238 in @xcite for a nonexpansive mapping @xmath184 with different proof , provided that the sequence @xmath168 satisfies the conditions of step @xmath246 of algorithm @xmath185 .
as we see in the previous section , @xmath0-lipschitz - type condition @xmath350 depends on two positive parameters @xmath23 and @xmath24 . in some cases
, these parameters are unknown or difficult to approximate . to avoid this difficulty , using linesearch method
, we modify extragradient algorithm .
we prove strong convergence of this new algorithm without assuming the @xmath0-lipschitz - type condition .
linesearch method has a good efficiency in numerical tests . here
, we assume that bifunction @xmath351 satisfies in conditions @xmath336 , @xmath352 and @xmath353 and also in following condition which @xmath4 is nonempty , convex and closed of @xmath147-uniformly convex , uniformly smooth banach space @xmath5 and @xmath354 is an open convex set containing @xmath4 , + ( @xmath355 ) @xmath14 is jointly weakly continuous on @xmath356 , i.e. , if @xmath175 and @xmath176 and @xmath177 are two sequences in @xmath354 converging weakly to @xmath35 and @xmath30 , respectively , then @xmath357 + * algorithm @xmath147 * step 0 .
: : let @xmath358 , @xmath359 and suppose that @xmath186 $ ] for some @xmath360 , @xmath188 $ ] for some @xmath361 , @xmath362 $ ] , where @xmath363 and @xmath364 , where @xmath365 ( @xmath366 ) is the @xmath367uniformly convexity constant of @xmath5 .
: : let @xmath193
. set @xmath368 step 2 .
: : obtain the unique optimal solution @xmath194 by solving the following convex problem @xmath369 step 3 .
: : if @xmath199 , then set @xmath256 . otherwise + step 3.1 .
; ; find @xmath370 the smallest nonnegative integer such that @xmath371 step 3.2 . ; ; set @xmath372 , @xmath373 and go to step 4 .
: : choose @xmath374 and compute @xmath375 .
if @xmath376 , then @xmath377 and @xmath378 otherwise .
: : compute @xmath379 if @xmath199 and @xmath380 then stop : @xmath201 .
otherwise , go to step 6 .
: : compute @xmath202 , where @xmath205 @xmath206 step 7 . : : set n:=n+1 , and go to step 2 .
the following lemma shows that linesearch corresponding to @xmath195 and @xmath194 ( step 3.1 ) is well defined .
[ lem18 ] assume that @xmath199 for some @xmath381
. then 1 .
there exists a nonnegative integer @xmath370 such that the inequality in ( [ equa41 ] ) holds .
2 . @xmath382 3 .
@xmath383 suppose that @xmath209 .
assume towards a contradiction that for each nonnegative integer @xmath370 , @xmath384 where @xmath385 it is easy to see that @xmath386 as @xmath387 .
using condition @xmath388 , we obtain @xmath389 since @xmath390 , letting @xmath387 in inequality ( [ equa15 ] ) , we get @xmath391 because of @xmath194 is a solution of the convex optimization problem ( [ equa38 ] ) , we deduce @xmath392 for all @xmath27 .
if @xmath393 , then @xmath394 it follows from inequalities ( [ equa16 ] ) and ( [ equa40 ] ) that @xmath395 therefore from inequality ( [ equa39 ] ) , we obtain @xmath396 since @xmath397 .
it follows from ( [ equa21 ] ) that @xmath398 , because of @xmath376 .
thus , @xmath399 or @xmath400 where contradict the assumption @xmath358 .
so , @xmath401 is proved .
now , we prove @xmath309 . since @xmath14 is convex , we obtain @xmath402 consequently from inequality ( [ equa42 ] ) , we get @xmath403\geq\frac{\alpha\rho_{n}}{\lambda_{n}}\phi(y_{n},x_{n})>0,\ ] ] because of @xmath376 .
therefore @xmath382 + the proof @xmath404 can be found in @xcite ( lemma @xmath405 ) .
if @xmath5 is a real hilbert space , then lemma [ lem18 ] is reduced to proposition @xmath406 in @xcite when @xmath407 we examine the stopping condition in the following lemma where its proof is similar to the proof lemma [ lem6 ] .
[ lem17 ] let @xmath199 , then @xmath251 . if @xmath199 and @xmath200 , then @xmath408 and @xmath201 .
if @xmath5 is a real hilbert space , then lemma [ lem17 ] is the same proposition @xmath409 in @xcite with different proof , provided that the sequence @xmath168 satisfies the conditions of step @xmath246 of algorithm @xmath147 .
[ lem14 ] suppose that @xmath351 is a bifunction satisfying conditions @xmath388 and @xmath353 .
let @xmath176 and @xmath311 be two sequences in @xmath354 such that suppose @xmath410 and @xmath411 , where @xmath412 .
then we have @xmath413 let @xmath414 , it follows from condition @xmath353 and the definition of @xmath415 that @xmath416 for all @xmath417 . taking the limits as @xmath323 , using the condition @xmath388 , we give @xmath418 for all @xmath417 .
hence , @xmath419 .
now , we prove the following proposition for algorithm @xmath147 , which have important role in the proof of main result in this section .
[ lem9 ] for all @xmath247 and all @xmath209 , we get 1
. @xmath420 2 .
@xmath421 using lemma [ lem4 ] , the definition of @xmath422 and inequality ( [ eq5 ] ) , we have @xmath423 since @xmath424 , we get @xmath425 therefore , we obtain @xmath426 on the other hand , from lemma [ eq6 ] , we get @xmath427 where @xmath428 is the @xmath147-uniformly convex constant of @xmath5 .
thus , combining inequalities ( [ equa24 ] ) and ( [ equa25 ] ) , we can derive @xmath401 . a similar argument as in the proof of lemma [ lem3 ] shows that @xmath429 using ( i ) , we see that @xmath430 therefore @xmath309 is proved .
[ thm1 ] assume that @xmath267 is a relatively nonexpansive mapping such that @xmath268 then the sequences @xmath269 , @xmath270 , @xmath271 , @xmath431 and @xmath272 generated by algorithm @xmath147 converge strongly to the some solution @xmath273 , where @xmath274 and @xmath275 is the sunny generalized nonexpansive retraction from @xmath5 onto @xmath276 .
let @xmath285 .
similar to the proof of theorem [ eq25 ] , we have @xmath432 which imply that @xmath176 and consequently @xmath305 converge strongly to @xmath433 and @xmath434 since @xmath435 , it follows from lemma [ lem9 ] @xmath309 that @xmath436 now , we prove that @xmath177 , @xmath311 and @xmath437 are bounded and @xmath438 , as @xmath323 .
suppose that @xmath439 since @xmath0 is lower semicontinuous respect to the first argument @xmath30 , from lemma [ lem8 ] , we deduce @xmath440 let @xmath441 , @xmath442 and @xmath443 , we have @xmath444 @xmath445 set @xmath446 in inequality ( [ equa5 ] ) and @xmath447 in inequality ( [ equa7 ] ) , we get @xmath448 on the other hand , we have @xmath449 therefore , using lemma [ lem16 ] , there exists @xmath450 such that @xmath451 from inequalities ( [ equa4 ] ) and ( [ equa13 ] ) , we obtain @xmath452 for all @xmath453 , put @xmath454 , where @xmath455 .
so @xmath456 for all @xmath457 .
therefore it follows from inequality ( [ equa26 ] ) that @xmath458 for all @xmath459 , all @xmath460 and @xmath461 , this means @xmath462 is multivalued monotone .
using lemma [ lem7 ] and lemma [ lem8 ] , we have @xmath463 if only if @xmath464 which implies that @xmath465 , thus @xmath466 for all @xmath27 and all @xmath467 . replacing @xmath468 and @xmath469 by @xmath195 and @xmath194 in inequality ( [ equa26 ] ) , respectively and interchanging @xmath30 by @xmath195 in inequality ( [ equa27 ] )
, we have @xmath470 for all @xmath471 and all @xmath467 .
so , @xmath472 , for all @xmath471 and since @xmath473 , we obtain @xmath474 since @xmath475 , it follows from lemma [ lem14 ] that @xmath476 thus , we can deduce from lemma [ lem20 ] that @xmath477 is bounded , because of @xmath478 .
so , from inequality ( [ equa28 ] ) , we obtain @xmath479 therefore , we can conclude that @xmath177 is bounded .
since @xmath311 is a convex combination of @xmath176 and @xmath177 , it is also bounded , hence , there exists a subsequence of @xmath311 , again denoted by @xmath311 , which converges weakly to @xmath480 . in a similar way , it follows from lemmas [ lem20 ] and [ lem14 ] that the sequence @xmath437 is bounded .
+ if @xmath481 then we have @xmath482 and @xmath378 and if @xmath483 , by the definition of @xmath484 , we obtain @xmath485 since @xmath486 and @xmath364 .
now , we show that @xmath334 and @xmath487 .
if @xmath199 then it follows from lemma [ lem1 ] ( i ) , that @xmath488 for all @xmath489 by letting @xmath323 in inequality ( [ equa44 ] ) and using condition @xmath388 , we get @xmath490 , because of @xmath491 , i.e , @xmath492 now , we let that @xmath376 , since @xmath493 is convex , we obtain @xmath494 therefore , @xmath495\leq f(z_{n},x_{n})\rightarrow0,\ ] ] as @xmath323 . by the step @xmath238 of algorithm @xmath147 and inequality ( [ equa45 ] ) , we have @xmath496\leq f(z_{n},x_{n})\rightarrow0.\ ] ] now , we consider two cases : + * case @xmath185 : * @xmath497 .
+ in this case , there exists @xmath498 and a subsequence of @xmath499 , again denoted by @xmath499 , such that @xmath500 and since @xmath491 , from inequality ( [ equa46 ] ) , we can conclude that @xmath501 thus , from lemma [ lem5 ] , we have @xmath502 which implies @xmath503 + * case @xmath147 : * @xmath504 .
+ let @xmath370 be the smallest nonnegative integer such that the step 3.1 of algorithm @xmath147 is satisfied , i.e. , @xmath505 where @xmath506 .
so , @xmath507 on the other hand , setting @xmath393 in lemma [ lem1 ] ( i ) , condition @xmath336 and equality ( [ equa6 ] ) imply that @xmath508 therefore , @xmath509 from inequalities ( [ equa47 ] ) and ( [ equa48 ] ) , we have @xmath510 taking the limits as @xmath323 in above inequality , we obtain @xmath511 , since @xmath512 . because of @xmath177 is bounded , there exists a subsequence of @xmath177 , again denoted by @xmath177 , which converges weakly to @xmath513 . by letting @xmath514 in inequality ( [ equa49 ] ) and using conditions
@xmath336 and @xmath388 , we get @xmath515 which implies that @xmath516 , because of @xmath517 .
so , if we take the limits as @xmath323 in inequality ( [ equa48 ] ) , then we can conclude that @xmath518 thus , from lemma [ lem5 ] , we have @xmath519 , which implies that @xmath520 . by lemma
[ lem1 ] , we have @xmath521\geq\langle jy_{n}-jx_{n},y_{n}-y\rangle,\ ] ] for all @xmath27 . by letting @xmath323 in inequality ( [ equa50 ] )
, it follows that @xmath522 , because of @xmath491 , this means @xmath492 now , we show that @xmath523 .
let @xmath524 and @xmath525 .
using lemma [ lem15 ] , there exists a continuous , strictly increasing and convex function@xmath315\rightarrow\mathbb{r}$ ] with @xmath316 such that for @xmath285 , we get @xmath526 and in a similar way , there exists a continuous , strictly increasing and convex function @xmath318\rightarrow\mathbb{r}$ ] with @xmath319 such that for @xmath285 , we obtain @xmath527 it follows from inequalities ( [ eq.n1 ] ) and ( [ eq.n2 ] ) that @xmath528 @xmath529 taking the limits as @xmath323 in inequalities ( [ equa30 ] ) and ( [ equa31 ] ) , using lemma [ lem10 ] and equality ( [ equa43 ] ) , we obtain @xmath530 from the properties of @xmath325 and @xmath326 , we have @xmath531 since @xmath257 is uniformly norm - to - norm continuous on bounded sets , we obtain @xmath532 so , we get @xmath341 , because of @xmath533 , therefore using the definition of @xmath184 , we have that @xmath342 setting @xmath343 in lemma [ lem4 ] , since @xmath202 and @xmath0 is continuous respect to the first argument , we obtain @xmath534 also , since @xmath347 , it follows from lemma [ eq38 ] that @xmath345 for all @xmath535 . therefore @xmath348 and consequently sequences @xmath269 , @xmath270 , @xmath271 , @xmath431 and @xmath272 converge strongly to @xmath349 .
now , we demonstrate theorems [ eq25 ] and [ thm1 ] with an example . also , we compare the behavior of the sequence @xmath176 generated by algorithms @xmath185 and @xmath147 . 1
. @xmath539 for all @xmath8 , 2 . if @xmath540 , then @xmath541 for all @xmath542 , i.e. , @xmath14 is pseudomonotone , while @xmath14 is not monotone .
3 . if @xmath543 and @xmath544 , then @xmath545 i.e. , @xmath14 is jointly weakly continuous on @xmath15 .
4 . let @xmath546 .
since @xmath547 so @xmath179 is convex , also , @xmath548 , hence @xmath179 is lower semicontinuous .
since @xmath549 , thus @xmath179 is subdifferentiable on @xmath4 for each @xmath550 5 . since @xmath551 , we get @xmath552 i.e. , @xmath14 satisfies the @xmath0-lipschitz - type condition with @xmath553 thus , @xmath560 , i.e. , @xmath184 is relatively nonexpansive mapping . on the other hand , if for each @xmath27 , @xmath561 , then @xmath562 , i.e. , @xmath563 and consequently @xmath564 also , assume that @xmath565 , @xmath566 for all @xmath278 and @xmath567 . in linesearch algorithm , @xmath195 is the same in extragradient algorithm .
assume that @xmath578 , @xmath579 , @xmath580 , @xmath581 and @xmath582 .
so , @xmath583 i.e. , @xmath584 and @xmath370 is the smallest nonnegative integer such that @xmath585 where @xmath586 also , @xmath587 and @xmath588 . since @xmath376 , then @xmath589 and @xmath590.\ ] ] furthermore , numerical results for algorithm @xmath147 show that the sequences @xmath176 , @xmath177 , @xmath311 , @xmath305 and @xmath591 converge strongly to @xmath246 . by comparing figure1 and figure2 ,
we see that the speed of convergence of the sequence @xmath176 generated by linesearch algorithm is equal to extragradient algorithm . the computations associated with example
were performed using matlab ( step:@xmath592 ) software .
j. p. aubin , _ optima and equilibria _ , springer , new york , ( 1998 ) .
e. blum and w. oettli , _ from optimization and variational inequality to equilibrum problems _ , the mathematics student , * 63 * ( 1994 ) , 127149 .
y. c. liou , _ shrinking projection method of proximal - type for a generalized equilibrium problem , a maximal monotone operator and a pair of relatively nonexpansive mappings _ , taiwanese j. math .
, * 14 * ( 2010 ) , 517540 . | in this paper , using sunny generalized nonexpansive retraction , we propose new extragradient and linesearch algorithms for finding a common element of the set of solutions of an equilibrium problem and the set of fixed points of a relatively nonexpansive mapping in banach spaces . to prove strong convergence of iterates in the extragradient method
, we introduce a @xmath0-lipschitz - type condition and assume that the equilibrium bifunction satisfies in this condition .
this condition is unnecessary when the linesearch method is used instead of the extragradient method .
a numerical example is given to illustrate the usability of our results .
our results generalize , extend and enrich some existing results in the literature .
[ multiblock footnote omitted ] = 15.8pt [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] |
99 korol a v , solovyov a v , and greiner w 1998 _ j.phys.g . : nucl .
phys . _ * 24 * l45 korol a v , solovyov a v , and greiner w 1999 _ int
_ e * 8 * 49 mikkelsen u and uggerhj e 2000 _ nucl .
inst . and meth .
_ b * 160 * 435 kumakhov m a 1976 _ phys .
_ a*57 * 17 zhang q - r 1999 _ int .
_ e * 8 * 493 madey j m j 1971 _ j.appl.phys .
_ * 42 * 1906 korol a v , solovyov a v , and greiner w 2000 _ int
_ e * 9 * 77 taratin a m and vorobiev s a 1988 _ nucl .
inst . and meth .
_ b * 31 * 551 ; 1989 _ nucl .
inst . and meth .
_ b * 42 * 41 arutyunov v a , kudryashov n a , samsonov v m , and strikhanov m n 1991 _ zh . tehn .
fiz . _ * 61 * 32 ; 1991 _ zh . tehn .
fiz . _ * 61 * 1 ; 1991 _ nucl .
_ b * 363 * 283 krause w , korol a v , solovyov a v , and greiner w 2000 to be submitted to _ phys .
e baier v n and katkov v m 1967 _ zh . eksp .
_ * 53 * 1478 ( engl .
1968 _ sov .
jetp _ * 26 * 854 ) ; 1968 _ zh . eksp .
fiz . _ * 55 * 1542 ( engl .
1969 _ sov .
jetp _ * 28 * 807 ) baier v n , katkov v m , and strakhovenko v m 1998 _ high energy electromagnetic processes in oriented single crystals _
( singapore : world scientific ) berestetskii v b , lifshitz e m , and pitaevskii l p 1982 _ quantum electrodynamics _ ( oxford : pergamon ) tsyganov e n 1976 _ fermilab preprint _
tm-682 , 684 ( batavia ) gemmell d s 1974 _ rev .
* 46 * 129 biruykov v m , chesnokov y a , and kotov v i 1996 _ crystal channelling and its application at high - energy accelerators _
( berlin : springer ) komarov f f 1979 _ phys .
( b ) * 96 * 555 sternheimer r m 1966 _ phys .
* 145 * 247 mason w p 1972 acoustic properties of solids _ american institute of physics handbook _
3rd edn ( new york : mcgraw - hill ) lindhard j 1965 _ kong .
danske vid .
selsk . mat .-
* 34 * 14 uggerhj e 1993 _ radiation effects and defects in solids _ * 25 * 3 kumakhov m a and komarov f f 1989 _ radiation from charged particles in solids _
( new york : american institute of physics ) | we present the results of numerical calculations of the channelling and undulator radiation generated by an ultra - relativistic positron channelling along a crystal plane , which is periodically bent
. the bending might be due either to the propagation of a transverse acoustic wave through the crystal , or due to the static strain as it occurs in superlattices .
the periodically bent crystal serves as an undulator .
we investigate the dependence of the intensities of both the ordinary channelling and the undulator radiations on the parameters of the periodically bent channel with simultaneous account for the dechannelling effect of the positrons .
we demonstrate that there is a range of parameters in which the undulator radiation dominates over the channelling one and the characteristic frequencies of both types of radiation are well separated .
this result is important , because the undulator radiation can be used to create a tunable source of x - ray and @xmath0-radiation . in this letter we present for the first time the results of calculations of the total spectrum of emitted photons accompanying the channelling process of ultra - relativistic charged particles through a crystal
which is periodically bent .
the present consideration is a further step in investigating new phenomenon which was described recently in @xcite and was called acoustically induced radiation ( air ) .
it was noted that the periodic pattern of crystal bendings ( which can be achieved either through propagation of a transverse acoustic wave , aw , or by using static periodically strained crystalline structures @xcite ) gives rise to a new mechanism of electromagnetic emission of the undulator type , in addition to a well - known ordinary channelling radiation @xcite . without any loss of generality
further in this letter we consider the case of a dynamic periodic bending of a crystal by means of transverse aw , as it is illustrated in . under the action of the transverse aw propagating along the @xmath1-direction , which defines the center line of initially straight channel ( not plotted in the figure )
the channel becomes periodically bent .
provided certain conditions are fulfilled @xcite the beam of particles , which enters the crystal at a small incident angle with respect to the curved crystallographic plane , will penetrate through the crystal following the bendings of its channel .
this results in transverse oscillations of the beam particles ( _ additional _ to the oscillations inside the channel due to the action of the interplanar force ) .
these oscillations become an effective source of spontaneous radiation of undulator type due to the constructive interference of the photons emitted from similar parts of the trajectory .
it was demonstrated @xcite that the system `` ultra - relativistic charged particle + periodically bent crystal '' serves as a new type of undulator , and , consequently , as a new source of undulator radiation of high intensity , monochromaticity and of a particular pattern of the angular - frequency distribution . as it was pointed out in @xcite
this scheme leads , in addition to the spontaneous radiation , to a possibility to generate stimulated emission , similar to the one known for a free electron laser @xcite in which the periodicity of a trajectory of an ultra - relativistic projectile is achieved by applying spatially periodic magnetic field . in connection with the stimulated air it was noted in @xcite that to achieve noticeable degree of amplification one
has to operate with a positron bunch of a high volume density , which , nevertheless , turned out to be achievable in modern accelerators as discussed in @xcite .
the main subject of our previous studies @xcite was the characteristics of the undulator air itself . due to this reason we primarily investigate the case of a high - amplitude aw , @xmath2 with @xmath3 standing for the interplanar spacing , propagating through the crystal .
less attention , except for some qualitative estimates , has been paid to the detailed investigation of the mutual influence of two types of radiation , the ordinary channelling radiation and the air , in forming the total spectrum of the radiation emitted by a channeled positron .
this topic is addressed in the present letter where we report our first quantitative results on numerical calculation of the spectrum of the emitted photons with both mechanisms taken into account simultaneously .
we demonstrate that there are ranges of ( i ) the parameters of aw , which are the amplitude , @xmath4 , the wavelength @xmath5 , and the sound velocity @xmath6 , ( ii ) the energies of projectile positron , @xmath7 , ( iii ) the crystal parameters , which include the length of a crystal , the constituent atoms and the types of crystallographic planes , inside which * the characteristic frequencies of the air and the ordinary channelling radiation are well separated , * the intensity of air is essentially higher than that of the ordinary channelling radiation , * the radiative spectrum is stable towards the total losses of the particle ( in the case of a positron it is primarily the radiative ones ) these items , except the last one which was considered in @xcite , are discussed below in the letter . to conclude the introductory part we mention that in the past the problem of evaluating the total spectrum of radiation formed in a bent crystal was considered in several publications @xcite in the case of a projectile channelling in a crystal bent with a constant curvature radius . in our letter for the first we investigate the problem for the periodically bent channel .
in full the results of our research will be published elsewhere @xcite . below
we present the essential points .
an adequate approach to the problem of the radiation emission by an ultra - relativistic particle moving in an external field was developed by baier and katkov in the late 1960s @xcite and was called `` operator quasi - classical method '' by the authors .
the details of that formalism can be found in @xcite .
the advantage of this method is that it allows to use the classical trajectory for the particle in an external field and , simultaneously , it takes into account the effect of the radiative recoil . for particles with spin @xmath8
the energy radiated into a given direction @xmath9 summed over the polarizations of the photon and the projectile is given by ( the cgs system is adopted throughout the paper ) @xmath10 where @xmath11 is the fine structure constant , @xmath12 , and @xmath13 \left({\bf v}(t_1){\bf v}(t_2)/c^2 - 1\right)+u^2\gamma^{-2}\right\}/2 $ ] , where @xmath14 is the velocity of light , @xmath15 is the relativistic factor and @xmath16 .
the main advantage of the the quasi - classical approach is that to calculate the angular - spectral distribution of the radiation one only needs to know the time dependencies of classical radius - vector @xmath17 and the velocity @xmath18 of projectile . in connection with a positron channelling through a periodically bent crystal
the above relations impose , if applied directly , some restrictions on the projectile energy , on the parameters of the crystal ( channel ) and the aw .
the channelling process in a bent crystal takes place if the centrifugal force in the channel is less than the maximal force due to the interplanar field @xcite . for a periodically bent crystal
the maximal centrifugal force is equal to @xmath19 , with @xmath20 and @xmath21 being a minimum curvature radius of the bent channel .
hence , the following condition must be fulfilled @xcite @xmath22 where @xmath23 stands for the maximum gradient of the interplanar field . for an acoustically bent channel @xmath24 ,
therefore , the inequality ( [ 1 ] ) bounds together the characteristics of the crystal , @xmath23 , the aw amplitude and the wavelength , and the relativistic factor @xmath0 . a comprehensive study of the allowed ranges of all parameters involved in ( [ 1 ] ) was carried out in @xcite .
second requirement which has to be fulfilled to make eq .
( [ wkb_1 ] ) directly applicable to calculating the spectrum accompanying the channelling process concerns the upper limit of integration @xmath25 , which is related to the crystal length @xmath26 through @xmath27 , and which , for given value of the aw wavelength , defines the number of the undulator periods as @xmath28 .
the air acquires specific features of the undulator - type radiation ( such as the monochromaticity and particular pattern of the angular - frequency distribution , see e.g. @xcite ) provided @xmath29 .
the length of a crystal is subject to a physical condition that the positron bunch stays inside the channel when penetrating through the crystal on the scale of @xmath26 .
in reality , the parasitic effect , the dechannelling leads to a decrease in the volume density of the channeled particles @xmath30 with penetration distance @xmath1 , and roughly satisfies the exponential decay law for both straight and bent channels ( see @xcite ) , @xmath31 , where @xmath32 is the volume density t the entrance , and @xmath33 is the dechannelling length , which for given crystal and channel depends on a positron energy and on the curvature radius . for a periodically bent crystal @xmath34
can be estimated as @xcite @xmath35 where @xmath36 is the classical radius of an electron , @xmath37 ( @xmath38 is the bohr radius ) and @xmath39 ev are , respectively , the thomas - fermi radius and ionization potential of the crystal atoms , @xmath40 is the atomic number , @xmath3 is the interplanar distance in the lattice .
the quantity @xmath41 is the critical ( minimal ) radius consistent with the channelling condition in a bent crystal ( [ 1 ] ) .
the quantity @xmath42 is the coulomb logarithm characterizing the ionization losses of an ultra - relativistic positron in amorphous media with account for the density effect ( see e.g. @xcite ) .
the air will have a pronounced pattern of undulator - type radiation provided the number of the periods is large on the scale of @xmath33 . combined with ( [ stim9 ] ) this condition leads to another restriction on the parameters involved @xmath43 where @xmath5 can be related to the aw frequency @xmath44 through @xmath45 with @xmath6 standing for the sound velocity .
illustrates the restrictions which are imposed on the values of @xmath4 and @xmath44 by inequalities ( [ 1 ] ) and ( [ 2 ] ) in the case of @xmath46 gev planar channelling in @xmath47 along ( 110 ) crystallographic planes .
the diagonal straight lines correspond to various values ( as indicated ) of the parameter @xmath48 consistent with the channelling condition ( [ 1 ] ) .
the curved lines correspond to various values ( as indicated ) of the number of undulator periods @xmath49 related to the dechannelling length @xmath33 through eq .
( [ 2 ] ) .
the horizontal lines mark the values of the aw amplitude equal to @xmath3 ( with @xmath50 being the ( 110 ) interplanar distance in @xmath47 ) and to @xmath51 .
the vertical line marks the value @xmath52 mhz of the aw frequency for which the spectra ( [ wkb_1 ] ) were calculated .
we used the value @xmath53 cm / s for the velocity of sound in @xmath47 ( this value was obtained by using the data from @xcite ) .
thus , the aw wave - length used in our calculations equals to @xmath54 cm .
the calculated spectra of the radiation emitted in the forward direction ( with respect to the @xmath1-axis , see ) for photon energies from 45 kev to 1.5 mev are presented in figures [ fig.3 ] .
the details of the analytical evaluation of the right - hand side of ( [ wkb_1 ] ) and its numerical implementation as well as more extended results of the calculation will be published soon @xcite . here
we briefly sketch the numeric procedure and present a discussion of the exhibited results .
the aw frequency , the number of undulator periods and crystal length were fixed at @xmath55 mhz , @xmath56 and @xmath57 cm .
the ratio @xmath58 of the aw amplitude to the interplanar spacing was varied within the interval @xmath59 .
the case @xmath60 corresponds to the straight channel . to evaluate the spectral distribution ( [ wkb_1 ] ) the following procedure was adopted .
firstly , the spectrum was calculated for individual trajectories .
these were obtained by solving the relativistic equations of motion with both the interplanar and the centrifugal potentials taken into account .
we used the continuum approximation @xcite to describe the projectile positron
lattice atoms interaction . within these scheme
we considered two frequently used @xcite analytic forms for the interplanar potential , the harmonic and the molire potentials calculated at the temperature @xmath61 k to account for the thermal vibrations of the lattice atoms .
for each @xmath58 value by changing the initial values of the entrance coordinate @xmath62 and the initial velocities along the @xmath63-axis ( see ) we left only those trajectories which corresponded to the case of stable channelling through the whole crystal length @xmath26 .
we call a trajectory as a `` stable '' one if moving along it the particle does not approach crystalline planes at a distance less than the thomas - fermi radius @xmath64 ( @xmath65 for a @xmath47 atom ) .
this allowed us to totally disregard , at least on the scale @xmath66 , the random scattering of a projectile by lattice electrons ( see e.g. @xcite ) .
thus , for each @xmath58 value , we defined the ranges of the initial coordinates @xmath67 $ ] and the velocities @xmath68 , and , correspondingly , the initial phase volume @xmath69 ( where @xmath70 stands for initial transverse momentum ) for which the corresponding classical trajectories are stable when channelling through the whole crystal length @xmath26 .
then , discretizing the found initial phase volume @xmath71 by choosing @xmath72 points @xmath73 , the individual spectra @xmath74 were calculated for each pair of the initial coordinate and velocity .
finally , for each @xmath58 value we evaluated the averaged spectrum defined as follows : @xmath75\ , p_y^{(0 ) } \d y^{(0 ) } \label{3}\ ] ] here , the integration is carried out over the phase volume @xmath71 , and the integral is scaled by the phase volume @xmath76 of stable trajectories in straight channel .
the ratio @xmath77 describes the number of particles channelled through the bent crystal relative to the number of particles channelled through the straight one .
hence , it is convenient to use the quantity @xmath78 to compare the spectra produced by effectively different number of projectiles as it happens for different @xmath58 values .
figures [ fig.3 ] correspond to the spectra ( [ 3 ] ) obtained as outlined above .
the results presented were calculated by using the molire approximation for interplanar potential .
the first graph in corresponds to the case of zero amplitude aw ( the ratio @xmath60 ) and , hence , presents the spectral dependence of the ordinary channelling radiation only .
the asymmetric shape of the calculated ordinary channelling radiation peak , which is due to the strong anharmonic character of the molire potential , bears close resemblance with the experimentally measured spectra @xcite .
the spectrum starts at @xmath79 kev , reaches its maximum value at @xmath80 kev , and steeply cuts off at @xmath81 kev .
this peak corresponds to the radiation into the first harmonic of the ordinary channelling radiation ( see e.g. @xcite ) , and there is almost no radiation into higher harmonics .
the latter fact is consistent with general theory of dipole radiation by ultra - relativistic particles undergoing quasi - periodic motion ( see e.g. @xcite ) .
dipole approximation is valid provided the corresponding undulator parameter @xmath82 is much less than 1 . in this relation @xmath83 and
@xmath84 stand for the characteristic scales of , correspondingly , the amplitude and the wave - length of the quasi - periodic trajectory . for the channelling motion one can estimate @xmath83 as @xmath85 , and @xmath86 , where @xmath87 standing for the characteristic period of the channelling oscillations ( using the harmonic approximation for the interplanar potential one gets @xmath88 where @xmath89 is the depth of the interplanar potential well ) . in the case of 0.5 gev positron channeled along ( 110 ) planes in @xmath47 one has @xmath90 , @xmath91 ev , @xmath92 @xcite . hence , @xmath93 and all the channelling radiation is concentrated within some interval in the vicinity of the energy of the first harmonic .
the latter one can estimate as ( see e.g. @xcite ) @xmath94 arriving at the value @xmath95 kev which is exactly the calculated maximum value @xmath80 kev .
more accurate estimates ( the details are presented in @xcite ) , based on the account for the dependence of the channeling oscillation period @xmath96 on the amplitude @xmath83 of the oscillations , also reproduce the calculated position and the width of the peak of the channeling radiation .
increasing the @xmath58 ratio leads to the modifications in the spectrum of radiation .
the changes which occur manifest themselves via three main features , ( i ) the lowering of the ordinary channelling radiation peak , ( ii ) the gradual increase of the intensity of undulator radiation due to the crystal bending , ( iii ) the appearing of additional structure ( the sub - peaks ) in the vicinity of the first harmonic of the ordinary channelling radiation .
let us discuss all these features .
the decrease in the intensity of the ordinary channelling radiation with the increase of the @xmath58 ratio is related to the simple fact that the growth of the aw amplitude leads to lowering of the allowed maximum value of the channelling oscillations amplitude @xmath83 ( this are measured with respect to the centerline of the bent channel ) @xcite . indeed , inside the bent channel the motion of the particle is subject to the action of the effective potential @xmath97 where @xmath98 $ ] is the ( local ) distance from the centerline and @xmath99 is the ( local ) curvature radius of the channel . for a channel bent by a transverse harmonic aw @xmath100\,\sin(2\pi z/\lambda)$ ] .
the particle could be trapped into the channelling mode provided its total energy associated with the transverse motion is less the minimal value , @xmath101 , of the two maxima points of the asymmetric potential well described by ( [ 2.25 ] ) @xcite .
the potential @xmath102 reaches the magnitude of @xmath101 at some point @xmath103 which satisfies the condition @xmath104 and the absolute value of @xmath103 decreases with the growth of @xmath4 .
hence , the larger the channel is bent the lower the allowed values of the channelling oscillations amplitude are , and , consequently , the less intensive is the channelling radiation , which is , essentially , proportional to the squared amplitude @xcite .
let us note , that since we have restricted the range of the @xmath83 values by imposing the condition @xmath105 ( see the discussion above ) , then the decrease of the intensity of channelling radiation occurs starting with some non - zero value of the aw amplitude for which @xmath106 also becomes less than @xmath107 .
these arguments explain the fact that the intensities of the ordinary channelling radiation for @xmath60 and @xmath108 are much alike , while for @xmath109 it starts loosing the magnitude .
the undulator radiation ( the air ) related to the motion of the particle along the centerline of periodically bent channel is absent in the case of straight channel ( the graph @xmath60 ) , and is almost invisible for comparatively small amplitudes of the aw ( see the graph for @xmath108 ) . with @xmath4 increasing
the peaks corresponding to air are becoming more prominent , and for large @xmath4 values ( @xmath110 ) two additional features appear : the intensity of the first harmonic of the air becomes larger than the intensity of the ordinary channelling radiation , and there appears radiation into the third harmonic of the air . the positions and the widths of the air peaks
can be quite accurately estimated as follows ( see @xcite ) : @xmath111 , where the integer @xmath112 enumerates the harmonics , and @xmath113 stands for the parameter of the undulator related to the periodicity of the channel bending .
the width of each peak @xmath114 is independent on @xmath115 .
substituting the values of @xmath0 , @xmath5 and @xmath3 indicated above one expresses the undulator parameter via the ratio @xmath58 : @xmath116 . even for the largest considered value @xmath117
the parameter @xmath118 is less than 1 , thus making the radiation into higher harmonics of the air almost negligible compared with the intensity radiated into the fundamental harmonic @xmath119 .
the latter is located at @xmath120 kev having the width @xmath121 kev which is almost 40 times less than the width of the peak of the channeling radiation .
these values depend neither on the ratio @xmath117 nor on the type of the interplanar potential . as mentioned , all graphs in refer to the forward emission .
therefore , in accordance with general theory of the undulator radiation ( see e.g. @xcite ) , the second peak of the air , which is mostly pronounced in the case @xmath117 , corresponds to the third harmonic of air , and is located at @xmath122 kev .
the intensities radiated into the fundamental and the third harmonics are equal to @xmath123 sr@xmath124 and @xmath125 , respectively .
their ratio is approximately equal to @xmath126 which is also in accordance with general theory .
the intensity of the air gradually increases with the aw amplitude .
more precisely , @xmath127 ( in the case @xmath128 ) , and this tendency one can observe comparing the air peaks in the graphs corresponding to @xmath129 . it is important to note that the positions of sharp air peaks , their narrow widths , and the radiated intensity are , practically , insensitive to the choice of the approximation used to describe the interplanar potential . in addition , provided the condition ( [ 1 ] ) is fulfilled , the air peaks are well separated ( in the photon energy scale ) from the peaks of the channeling radiation .
therefore , if being interested in the spectral distribution of the air only , one may disregard the channeling oscillations and to assume that the projectile moves along the centerline of the bent channel @xcite .
the above statements are illustrated by where we compare the results of calculations of the total spectrum ( [ 3 ] ) in vicinity of the first harmonic of air in the case @xmath117 .
all parameters are the same as in .
the filled and open circles represent the results of evaluation of the right - hand sides of ( [ 1 ] ) and ( [ 3 ] ) accompanied by numerical solution of the equations of motion for the projectile within the molire ( filled circles ) and the harmonic ( open circles ) approximations for the interplanar potential .
the solid line corresponds to the air radiation only ( in this case the numerical procedures are simplified considerably , leading to the reduction , by orders of magnitude , of the cpu time ) .
it is clearly seen that the more sophisticated treatment almost does not change the profile of the peak obtained by means of simple formulae describing purely air radiation @xcite .
moreover , the minor changes in the position and the height of the peak can be easily accounted for within the framework of the cited formalism by introducing the effective undulator parameter @xcite and ( in the case of the harmonic approximation ) the effective undulator amplitude @xcite .
thus , in vicinity of the air peaks there is no coupling of the two mechanisms of the radiation . on the contrary
, the air strongly affects the spectrum in the photon energy range corresponding to the channeling radiation .
the discussion of this effect lies beyond the scope of this letter and is presented in @xcite . here
we only mention that , as it is clearly seen from the graphs @xmath130 in , for large values of @xmath58 the peaks of the ordinary channeling radiation acquire additional structure : there appear sub - peaks separated ( in the case of the forward emission ) by the interval @xmath131 .
the presented results of the numerical calculations of the total spectrum of radiation formed in an acoustically bent crystal clearly demonstrate the validity of the statement made in @xcite that the air and the ordinary channeling radiation occur in essentially different ranges of the emitted photons energies , allowing , thus , to investigate the air properties separately from the ordinary channeling radiation .
the positions of the peaks of the air as well as their intensity can be easily varied by changing the energy of projectile , the crystal type , the type of crystallographic plane , and the parameters of periodic pattern of crystal bendings .
this statement by no means is restricted to the case of the acoustically bent crystal , but has more general nature .
the treatment and the main results presented above can be applied , in particular , to consider the undulator radiation in statically bent crystals @xcite .
whatever way is chosen to prepare periodically bent crystalline lattice it is worth investigating , both theoretically and experimentally , not only the spontaneous emission of the air - type , but the stimulated emission as well @xcite .
the authors express their gratitude to professor e. uggerhj for sending a copy of the paper @xcite prior to its publication .
we want to thank the administrative staff of the physics cip computer cluster at university of munich for the possibility of using some of their computers for numerical calculations .
the research was supported by dfg , gsi , and bmbf .
avk and avs acknowledge the support from the alexander von humboldt foundation . |
when the displacement of the pendulum is not restricted to small values , we are considering a ( general ) pendulum ( or oscillator ) . a pendulum oscillating ( or rotating ) about a rigidly fixed ( point )
fulcrum is exactly equivalent to a particle moving in a medium of friction coefficient @xmath0 along a potential @xmath1 and driven by the external periodic forcing @xmath2 .
the equation of motion of the pendulum ( identifying @xmath3 with the angular displacement @xmath4 , etc ) is given by : @xmath5 here the system is considered to be in equilibrium with an environment at temperature @xmath6 and the motion is underdamped @xmath7 where @xmath8 is the natural frequency of oscillation .
the driven pendulum , when the damping is small , has so far not found exact analytical description @xcite .
therefore , we resort to other , for example numerical and experimental , means of description .
the problem of pendulum has exact analogies in , and can be considered as the prototype of , many other phenomena of interest @xcite .
apart from a particle moving on the surface of a sinusoidal potential , some of the examples being ionic motion in a superionic conductor , and a phase motion in a josephson junction . the structure of the equations of motion for these phenomena is identical only the physical significance of the parameters differ .
the dimensionless form of the equation of motion is : @xmath9 here , all parameters are dimensionless and written by taking @xmath10 as independent parameters and setting @xmath11 .
note that , in these dimensionless units , the natural ( angular ) frequency @xmath12 turns out to be 1 . for this dimensionless equation of motion , even in the case of underdamped motion ,
the allowed space of parameters @xmath13 is vast .
however , for our purpose , we restrict our investigation to a limited range of @xmath14 space with @xmath15 .
note that if we view @xmath16 as a force derived from an effective potential @xmath17 , the potential , @xmath18 , gets tilted by an effective slope of @xmath19 . in the extreme cases , @xmath20 , so that the potential barrier disappears , momentarily or for a longer period , only for @xmath21 .
@xmath15 , thus corresponds to the potential barrier remaining finite throughout the period of @xmath22 .
that is , in the pendulum parlance , unless provoked by an appropriate additional forcing , for example , by thermal fluctuations , the pendulum can not go over the top and make a full circle .
of course , this important possibility of going over the potential barrier will also be considered as we proceed by including thermal fluctuations .
interesting numerical results , supported by experiment ( see figs . 1 and 2 and table 1 ) , are obtained in the range , @xmath23 and @xmath24 , of parameter space @xmath14 or @xmath25 , with @xmath26 . since the potential @xmath27 is periodic , the choice of initial positions @xmath28 , suffices to obtain all possible trajectories , @xmath29 , @xmath30 . of course , we just release the pendulum initially from a static condition @xmath31 ( at @xmath32)@xmath33 .
however , one could take the initial velocity @xmath34 non - zero too without altering the essential qualitative results of our interest .
frequency=@xmath35 & @xmath0 & @xmath36=@xmath0@xmath37 + no .
& dimensionless & hertz & dimensionless & @xmath38farad + 1&6.0&164.22&0.07 & 0.216457 + 2&6.5&151.59&0.08 & 0.247379 + 3&7.0&140.76&0.09 & 0.27830 + 4&7.5&131.38&0.10 & 0.30922 + 5&8.0&123.16&0.11 & 0.34015 + 6&8.5&115.92&0.12 & 0.37107 + 7&9.0&109.48&0.13 & 0.40199 + 8&9.5&103.72&0.14 & 0.43291 + 9&10.0&98.53&0.15 & 0.46384 + 10&10.5&93.84&0.16 & 0.49476 + 11&11.0&89.57&0.17 & 0.52568 +
the circuit shown in figure ( 1 ) simulates ( at a ) the equation : @xmath39 where we have taken @xmath40volt , @xmath41ampere with @xmath42volt .
eqn.(2.1 ) , is written in dimensionless units by setting @xmath43 , @xmath44 , @xmath45 .
the equation , with reduced variables denoted again now by the same symbols , corresponding to equation ( 2.1 ) is written as : @xmath46 this is to be compared with equation ( 1.2 ) .
here @xmath47 , @xmath48 . in our experiment ,
the output voltage @xmath49 represents the trajectory @xmath29 ( or the angular displacement @xmath50 ) of equation ( 1.2 ) .
we perform the experiment in the earlier stated region of parameter space @xmath25 i.e. , in the space of @xmath51 , and for @xmath52 or the amplitude @xmath53 near about 0.2 of the drive current @xmath54 .
figure 3 shows the result of our experiment .
it shows that for the same periodic drive , we obtain two ( and only two ) possible kinds of trajectories depending on the initial conditions taken .
one of them oscillating with a large amplitude and having a large phase lag ( figure 3 ( b ) ) with respect to the externally applied field @xmath54 and the other oscillating with a small amplitude and having a small phase lag ( figure 3 ( d ) ) . in other words , for a given applied field @xmath54 at @xmath55 one gets a fraction of large amplitude output signals and the remaining being of small amplitude . our experimental result , thus , verifies the existence of the two solutions or states of trajectories of a pendulum which was numerically discovered earlier .
of the output voltage signal as a function of the period @xmath56 of the drive for @xmath57.,width=566,height=302 ] ( a ) of the output voltage signal as a function of the period @xmath56 of the drive for @xmath57.,width=566,height=302 ] in fig . 4 and fig . 5
, we see the nature of the amplitude and phase lag of the output voltage signal as a function of @xmath56 for various values of @xmath0 . in fig .
4 , we plot the average amplitude @xmath58 , and in fig . 5
, we plot the phase lags @xmath59 between the states of output voltage signal @xmath60 and the drive field @xmath54 as @xmath61 is varied .
we notice that both a@xmath62 and @xmath59 change continuously as @xmath56 is varied for both states for @xmath57 .
we obtain two branches of the output voltage signal for @xmath63 .
plots at the top refers to the large amplitude state ( la ) having a large amplitude ( fig .
4 ) and a large phase lag ( fig .
5 ) , and the plots at the bottom refers to the small amplitude states ( sa ) having a small amplitude ( fig .
4 ) and a small phase lag ( fig . 5 ) .
there is a region of @xmath56 where both the two states coexist .
these region of @xmath56 where both the states overlap is shown in the inset of fig .
the plot in the inset shows that for small @xmath0 values , the range of @xmath56 values where both the two states coexist is large .
however , as @xmath0 value is increased , we see that the range of @xmath56 values decreases till it reaches a point where the two dynamical states terminate ( merge into one ) . the value of @xmath0 where the two dynamical states terminate is found to be @xmath64 . in this region where the two states terminate , we can see only one kind of output voltage signal where @xmath59 and a@xmath62 change continuously as @xmath56 is varied in the entire range .
therefore , one can not identify from the graph to call the output voltage signal sa or la for @xmath65 . when compared the range of @xmath56 values ( inset of fig .
4 ) where both the states overlapped with the numerical result ( fig .
6 ) , we can see that experimentally , the range of @xmath56 values is very small .
this is due to the fact that we have explore only few values of the initial condition .
an exhaustive initial conditions , however , may change the range of @xmath56 values by a small amount . showing the coexistence of the two dynamical states for different @xmath0 values ( see fig .
3 [ 8]).,width=566,height=302 ] numerically , it has been found that these two `` trajectories '' are distinct and robust and can be given the status of dynamical states having their own basins of attraction in the @xmath66 phase plane @xcite .
it will be appropriate to remark that the result is not only interesting it is novel as well .
the importance of the discovery of these dynamical states is enhanced by the fact that as the temperature is raised from zero ( as a noise signal is added to the input signal ) the two states make transitions between them . in other words , as additional provocations due to thermal fluctuations are included , the fraction of the large amplitude states ( and hence that of the small amplitude state ) changes .
we could observe this in our experiment by adding a feeble gaussian noise to the input signal .
numerically @xcite , it was found that @xcite , for @xmath15 , as the temperature is raised gradually from zero , initially large amplitude states ( from among a collection or ensemble of them ) begin to jump over to the small amplitude state and not the other way round .
thus , as the temperature is raised , the fraction of large - amplitude states gradually decrease and attain a minimum . but from then on , as the temperature is increased further , the small amplitude states too begin to jump over to the other state and the fraction of large amplitude states begins to rise from the minimum .
this process continues till a temperature at which the fraction becomes close to 0.5 .
however , as the temperature increases the distinction between the two states gets blurred and the transitions are so frequent that it is hard to observe a state over a significant fraction of a period of @xmath22 .
thus , in a sense , we have a bistable system separated by a barrier ; for @xmath15 , the large amplitude states faces a smaller barrier than the small amplitude state@xcite .
note that oblivious to the underlying sinusoidal ( periodic , multistable ) potential the motion of the pendulum is being described in the language of a bistable system consisting of the two dynamical states of trajectories . though quite imperfectly , one is tempted to draw analogy with the phonons ( normal modes of lattice vibrations or quasiparticles ) in a crystal at small nonzero temperatures .
just as one calculates the specific heat , for example , of crystals at low temperatures in terms of density of phonons , one can describe a phenomenon called stochastic resonance @xcite in a sinusoidal potential in terms of , for example , the fraction of large amplitude states .
it is to be reminded that the change in the fraction is brought about by the transition between the two dynamical states .
the explanation of the possibility of occurrence of stochastic resonance in a sinusoidal potential is thus clothed in terms of a `` bistable potential '' @xcite !
the occurrence of stochastic resonance is indicated by the peaking of the mean hysteresis loop area @xmath67 as a function of noise strength where @xmath29 is the trajectory of particles due to the application of the periodic drive @xmath22 .
this result was obtained numerically @xcite . in order to observe the occurrence of stochastic resonance in sinusoidal potentials we need in our experiment an additive gaussian white noise signal to be added to the input voltage signal of fig .
1 . we are in the process of implementing this to obtain stochastic resonance in our experiment from the current - voltage characteristic of the circuit as a function of noise strength .
note that so far the occurrence of stochastic resonance has been observed experimentally only in bistable systems @xcite .
our present effort will provide an experimental support to the numerical result obtained earlier on the occurrence of stochastic resonance in periodic potentials .
in this experiment , the operational amplifier that we have use is the ua741 .
if , instead of the ua741 , we use a better performance operational amplifier , we expect to see a relative change in the range of @xmath56 values for different @xmath0 values but the qualiative results will remain the same .
for example , if we use the operational amplifier ad797 , there is a relative increase in the range of @xmath56 values for different @xmath0 values . from fig .
( 3 ) , we see that two dynamical states exist experimentally .
stochastic resonance has been found earlier to occur due to the transition between these two dynamical states when temperature @xmath68 .
it will be interesting to see the response of the output voltage signal when @xmath68 whether the transition between the two dynamical states do really lead to stochastic resonance . in stochastic resonance ,
one calculates the hysteresis loop area per period of the drive field @xmath54 .
the relative abundance of these two dynamical states for different value of temperature decides the value of the average hysteresis loop area . in this experiment ,
the temperature is taken to be @xmath55 since we havent add a gaussian white noise to the input voltage signal .
the frequency ( or bandwitdh ) of the gaussian white noise has to be very much greater than the frequency of the input voltage signal .
in fact , numerically it is @xmath698000 times higher than the frequency of the input voltage signal .
99 c. kittel , w.d .
knight , and m.a .
ruderman , , - volume 1 , mcgraw - hill book company , inc .
, new york , ch . 7 , 1965 .
h. risken , , springer - verlag , ch . 11 , 1989 . s. saikia , a.m. jayannavar , and m.c .
mahato , stochastic resonance in periodic potentials , phys .
e * 83 * , 061121 ( 2011 ) .
experimentally , we have not gone beyond taking an external signal with very small noise .
large noise regime is yet to be explored .
reenbohn , and m.c .
mahato , relative stability of dynamical states and stochastic resonance in a sinusoidal potential , phys .
e * 88 * , 032143 ( 2013 ) .
m.c . mahato and a.m. jayannavar , stochastic resonance , physics education ( india )
* 15 * , 323 ( 1999 ) . for an early review see l. gammaitoni , p. hanggi , p. jung , and f. marchesoni , stochastic resonance , rev .
phys . * 70 * , 223 ( 1998 ) .
reenbohn , and m.c .
mahato , dynamical states , stochastic resonance , and ratchet effect in a biharmonically driven sinusoidal potential , phys .
e * 91 * , 052151 ( 2015 ) . | just as in a harmonic oscillator ( simple pendulum ) one finds a theoretical model to explain many phenomena in physics , a pendulum is also a prototype model for explaining many phenomena in periodic systems .
an underdamped pendulum when driven by an appropriate sinusoidal field is found to have two distinct solutions , one having a large amplitude and the other having a small amplitude depending on the initial position of release .
these two solutions or states of motion has been found earlier numerically .
we set up an analog simulation experiment to study the system and verify the numerical results .
the analog experiment can be used further to study the phenomenon of stochastic resonance in a periodic potential . |
in the mathematical investigations of ecological systems , conservative dynamics are often considered non - robust , thus unrealistic as a faithful description of reality @xcite . through recent studies of stochastic , nonlinear population dynamics , however
, a new perspective has emerged @xcite : the stationary behavior of a complex stochastic population dynamics almost always exhibits a divergence - free cyclic motion in its phase space , even when the corresponding system of differential equations has only stable , node - type fixed points @xcite . in particular
, it has been shown that an underlying volume preserving conservative dynamics is one of the essential keys to understand the long - time complexity of such stochastic systems @xcite .
the aim of the present work , following a proposal in @xcite , is to carry out a comprehesive stochastic dynamic and thermodynamic analysis of an ecological system with sustained oscillations . in the classical studies on statistical mechanics , developed by helmholtz , boltzmann , gibbs , and others ,
the dynamical foundation is a hamiltonian system @xcite .
the theory in @xcite generalized such an approach that requires no _ a priori _ identification of variables as position and momentum ; it also suggested a possible _ thermodynamic structure _ which is purely mathematical in nature , independent of newtonian particles . in the context of population dynamics , we shall show that the mathematical analysis yields a _ conservative ecology_. among ecological models , the lotka - volterra ( lv ) equation for predator - prey system has played an important pedagogical role @xcite , even though it is certainly not a realistic model for any engineering applications .
we choose this population system in the present work because its mathematics tractability , and its stochastic counterpart in terms of a birth and death process @xcite .
it can be rigorously shown that a smooth solusion to lv differential equation is the law of large numbers for the stochastic process @xcite . in biochemistry ,
the birth - death process for discrete , stochastic reactions corresponding to the mass - action kinetics has been called a delbrck - gillespie process @xcite .
the lv equation in non - dimensionalized form is @xcite : @xmath4 in which @xmath5 and @xmath6 represent the populations of a prey and its predator , each normalized with respect to its time - averaged mean populations . the @xmath7 term in @xmath8 stands for the rate of consumption of the prey by the predator , and the @xmath9 term in @xmath10 stands for the rate of prey - dependent predator growth . and @xmath11 , and @xmath12 .
right panel : with @xmath13 , and @xmath14 , from outside inward , all with @xmath15 .
we see that the larger the @xmath0 , the smaller the temporal variations in the prey population , relative to that of predator .
, title="fig:",width=268 ] and @xmath11 , and @xmath12 .
right panel : with @xmath13 , and @xmath14 , from outside inward , all with @xmath15 .
we see that the larger the @xmath0 , the smaller the temporal variations in the prey population , relative to that of predator . , title="fig:",width=268 ] it is easy to check that the solutions to ( [ theeqn ] ) in phase space are level curves of a scalar function @xcite @xmath16 we shall use @xmath17 to denote the solution curve @xmath18 , and @xmath19 to denote the domain encircled by the @xmath17 .
[ fig_1 ] shows the contours of @xmath20 with @xmath21 and @xmath15 with different @xmath0 s .
let @xmath22 be the period of the cyclic dynamics .
then it is easy to show that @xcite @xmath23 furthermore ( see appendix [ app - a ] ) , @xmath24 in which @xmath25 is the area of @xmath26 , encircled by @xmath17 , using lebesgue measure in the @xmath7-plane .
the appropriate measure for computing the area will be further discussed in sec .
the parameter @xmath0 represents the relative temporal variations , or dynamic ranges , in the two populations : the larger the @xmath0 , the greater the temporal variations and range in the predator population , and the smaller in the prey population .
the paper is organized as follows . in sec .
[ sec2 ] , an extended conservation law is recognized for the lotka - volterra system . then the relationship among three quantities : the energy " function @xmath20 , the area @xmath2 encircled by the level set @xmath17 , and the parameter @xmath0 is developed . according to the helmholtz theorem , the conjugate variables of @xmath2 and @xmath0 are found as time averages of certain functions of population @xmath5 and @xmath6 .
analysis on those novel
state variables " demonstrates the tendency of change in mean ecological quantities like population range or ecological activeness when the parameter @xmath0 or energy @xmath27 varies . in sec .
[ sec3 ] , we show that the area @xmath2 encircled by @xmath17 is related to the concept of entropy . in sec .
[ sec4 ] , the conservative dynamics is shown to be an integral part of the stochastic population dynamics , which necessarily has the same invariant density as the deterministic conservative dynamics . in the large population limit
, a separation of time scale between the fast cyclic motion on @xmath17 and the slow stochastic crossing of @xmath17 is observed in the stochastic dynamical system .
the paper concludes with a discussion in sec .
( [ theeqn ] ) is not a hamiltonian system , nor is it divergence - free @xmath28 it can be expressed , however , as @xmath29 with a scalar factor @xmath30 .
one can in fact understand this scalar factor as a `` local change of measure '' , or time @xmath31 @xcite : @xmath32 for @xmath33 where @xmath34 satisfies the corresponding hamiltonian system . in sec .
[ sec3 ] and [ sec4 ] below , we shall show that @xmath35 is an invariant density of the liouville equation for the deterministic dynamics ( [ theeqn ] ) , and more importantly the invariant density of the fokker - planck equation for the corresponding stochastic dynamics . as will be demonstrated in sec .
[ sec2_2 ] and [ sec2_3 ] , statistical average of quantities according to the invariant measure @xmath36 can be calculated through time average of those quantities along the system s instantaneous dynamics .
knowledge about the system s long term distribution is not needed during the calculation .
these facts make the @xmath35 the natural measure for computing area @xmath2 .
any function of @xmath20 , @xmath37 is conserved under the dynamics , as is guaranteed by the orthogonality between the vector field of ( [ theeqn ] ) and gradient @xmath38 @xcite : @xmath39 this is analogous to the conservation law " observed in hamiltonian systems .
the nonlinear dynamics in ( [ theeqn ] ) , therefore , introduces a `` conservative relation '' between the populations of predator and prey according to ( [ hxy ] ) .
if we call the value @xmath20 an `` energy '' , then the phase portrait in the left panel of fig .
[ fig_1 ] suggests that the entire phase space of the dynamical system is organized according to the value of @xmath27 . the deep insight contained in the work of helmholtz and boltzmann
@xcite is that such an energy - based organization can be further extended for different values of @xmath0 : therefore , the energy - based organization is no longer limited to a _
orbit , nor a _ single _ dynamical system ; but rather for the entire class of parametric dynamical systems . in the classical physics of newtonian mechanical energy conservation , this yields the mechanical basis of the fundamental thermodynamic relation as a form of the first law , which extends the notion of energy conservation far beyond mechanical systems @xcite .
more specifically , we see that the area @xmath2 in fig .
[ fig_1 ] , or in fact any geometric quantification of a closed orbit , is completely determined by the parameter @xmath0 and initial energy value @xmath40 .
therefore , there must exist a bivariate function @xmath41 , assuming the implicit function theorem applies , then one has @xmath42 note that in terms of the eq .
( [ hfunction ] ) , a `` state '' of the ecological system is not a single point @xmath43 which is continuously varying with time ; rather it reflects the geometry of an entire orbit
. then eq .
( [ hfunction ] ) implies that any such ecological state has an `` h - energy '' , _ if _ one recognizes a geometric , state variable @xmath2 .
( [ hfunction ] ) can be written in a differential form @xmath44 in which one first introduces the h - energy for an ecological system with fixed @xmath0 via the factor @xmath45 .
then , holding @xmath2 constant , one introduces an `` @xmath0-force '' corresponding to the parameter @xmath0 . in classical thermodynamics ,
the latter is known as an `` adiabatic '' process .
the helmholtz theorem expresses the two partial derivatives in ( [ dh ] ) in terms of the dynamics in eq .
( [ theeqn ] ) . for canonical hamiltonian systems ,
lebesgue measure is an invariant measure in the whole phase space . on the level set @xmath17
, the projection of the lebesgue measure , called the liouville measure , also defines an invariant measure on the sub - manifold .
if the dynamics on the invariant sub - manifold @xmath17 is ergodic , the average with respect to the liouville measure is equal to the time average along the trajectory starting from any initial condition @xmath46 satisfying @xmath47 . as we shall show below , the invariant measure for the lv system ( [ theeqn ] ) in the whole phase space is @xmath48 .
projection of this invariant measure onto the level set @xmath17 can be found by changing @xmath43 to intrinsic coordinates @xmath49 : @xmath50 where @xmath51 is the unit normal vector of the the level set @xmath17 ; and @xmath52 . noting that : @xmath53 we have @xmath54 that is : @xmath55 where @xmath56 is the projected invariant measure of the lotka - volterra system on the level set @xmath17 .
it is worth noting that @xmath57 on the level set @xmath17 .
since dynamics on @xmath17 is ergodic , the average of any function @xmath58 under the projected invariant measure on @xmath17 is equal to its time average over a period : @xmath59 under the invariant measure @xmath35 , the area @xmath2 encircled by the level curve @xmath17 is : @xmath60 using green s theorem the area @xmath61 can be simplified as @xmath62 where @xmath63 is the time period for the cyclic motion . furthermore , @xmath64 that is @xmath65 in which @xmath66 is the time average , or phase space average according to the invariant measure .
we can also find the derivative of the area @xmath61 encircled by the level curve @xmath17 with respect to the parameter of the system @xmath0 as : @xmath67 in this setting , the helmholtz theorem reads @xmath68 in which @xmath69 the factor @xmath70 here is the mean @xmath71 , or @xmath72 , precisely like the mean kinetic energy as the notion of temperature in classical physics , and the virial theorem .
the @xmath0-force is then defined as @xmath73 it is important to note that the definition of @xmath3 given in the right - hand - side of ( [ f - alpha ] ) is completely independent of the notion of @xmath2 , even though the relation ( [ dheq28 ] ) explicitly involves the latter .
@xmath74 is a function of both @xmath75 and @xmath0 , however .
therefore , the value of @xmath0-work @xmath76 depends on how @xmath75 is constrained : there are iso-@xmath75 processes , iso-@xmath1 processes , etc .
@xcite the notion of an equation of state first appeared in classic thermodynamics @xcite . from a modern dynamical systems standpoint ,
a fixed point as a function usually is continuously dependent upon the parameters in a mathematical model , except at bifurcation points .
let @xmath77 be a globally asymptotically attractive fixed point , and @xmath0 be a parameter , then the function @xmath78 constitutes an _ equation of state _ for the long - time `` equilibrium '' behavior of the dynamical system .
if a system has a globally asymptotically attractive limit set that is not a simple fixed point , then every geometric characteristic of the invariant manifold , say @xmath79 , will be a function of @xmath0 .
in this case , @xmath80 could well be considered as an equation of state .
an `` equilibrium state '' in this case is the entire invariant manifold .
the situation for a conservative dynamical system with center manifolds is quite different . in this case , the long - time behavior of the dynamical system , the foremost , is dependent upon its initial data .
an equation of state therefore is a functional relation among ( i ) geometric characteristics of a center manifold @xmath79 , ( ii ) parameter @xmath0 , and ( iii ) a new quantity , or quantities , that identifies a specific center manifold , @xmath75 .
this is the fundamental insight of the helmholtz theorem . in ecological terms , area under the invariant measure :
@xmath2 , gives a sense of total variation in both the predator s and the prey s populations .
therefore , @xmath81 measures population range of both populations as a whole .
the parameter @xmath0 , on the other hand , is the proportion of predators over preys population ranges of time variations : @xmath82 the new quantity @xmath1 can be viewed as a measure of the mean ecological `` activeness '' : @xmath83 it is the mean of distance " from the prey s and predator s populations @xmath84 and @xmath85 , to the fixed populations in equilibrium @xmath86 . for population dynamic variable @xmath87 , eq .
[ eq2700 ] suggests a norm @xmath88 .
then , @xmath89 ; and an averaged norm of per capita growth rates in the two species : @xmath90 and finally , @xmath91 is the ecological force " one needs to counteract in order to change @xmath0 .
in other words , when @xmath92 is greater , more @xmath75-energy change is needed to vary @xmath0 .
it is also worth noting that @xmath92 is positively related to the prey s average population range .
in fact we can define another distance " of the prey s population @xmath84 to @xmath93 as : @xmath94 , then @xmath95 .
note that for very small @xmath87 : @xmath96 , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] , @xmath97 , and @xmath98 in the top row , different views among @xmath99 in the second row , and among @xmath100 in the third row.,title="fig:",width=182 ] fig .
[ fig_2 ] shows various forms of `` the equation of state '' , e.g. , relationships among the triplets @xmath101 , @xmath102 , and @xmath103 in the first row ; among the triplet @xmath104 in the second row ; and among the triplet @xmath105 in the third row .
the second row shows that the relation among @xmath104 is just like that among @xmath106 in ideal gas model : mean ecological activeness @xmath1 increases nearly linearly with the ecological force @xmath107 for constant @xmath0 , and with the proportion @xmath0 of the predator s population range over the prey s , for constant @xmath107 ; ecological force @xmath107 and the proportion @xmath0 of population ranges are inversely related under constant mean activeness @xmath1 . and when @xmath108 , @xmath109 .
other features can be observed by looking into the details of each column . the first column of fig .
[ fig_2 ] demonstrates that as the proportion @xmath0 of population ranges increases , the ecological force @xmath107 is alleviated ( for given @xmath75-energy or ecological activeness @xmath1 ) .
this is due to the positive relationship between the ecological force @xmath107 and the prey s population range ( as shown in eq .
[ f_alpha ] ) . since @xmath0 is the proportion of the predator s population range over the prey s , @xmath107 and @xmath0
would be inversely related when any resource , @xmath75-energy , or activity , @xmath1 , remains constant .
this fact means that on an iso-@xmath75 or iso-@xmath1 curve , when the proportion @xmath0 is large , relatively less @xmath75-energy change is needed to reduce it .
the first column also demonstrates an inverse relationship between @xmath0 and the total population range @xmath81 for any given @xmath1 , which reflects the fact that as the proportion of the predator s population range over the prey s increase , the total population range of the two species would actually decrease .
the second column of fig .
[ fig_2 ] demonstrates that : the ecological force @xmath107 and the total population range @xmath81 increases with the mean activeness @xmath1 ( with given @xmath75-energy or @xmath0 ) .
this observation means that it would also take more @xmath75-energy to change the proportion @xmath0 of the predator s population range over the prey s , if mean ecological activeness rises , and that more population range would be explored with more ecological activeness @xmath1 .
the third column about the relation between @xmath1 and @xmath0 is interesting : under constant @xmath75-energy , as the proportion @xmath0 of population ranges increases , the ecological activeness @xmath1 decreases , in accordance with the drop in the total population range @xmath81 as shown in fig [ fig_1 ] .
but when the total population range @xmath81 or the ecological force @xmath3 is to remain constant , ecological activeness @xmath1 actually increases with @xmath0 .
this means that under constant resource ( @xmath75-energy ) , the proportion @xmath0 of the predator s population range over the prey s restricts mean ecological activeness .
but if we fix the ecological force or total population range ( supplying more @xmath75-energy ) , an increase in predator s population range over prey s can increase ecological activeness .
nonlinear dynamics described by eq .
( [ theeqn ] ) has a linear , first - order partial differential equation ( pde ) representation @xmath110 a solution to ( [ thepde ] ) can be obtained via the method of characteristics , exactly via ( [ theeqn ] ) .
( [ thepde ] ) sometime is called the liouville equation for the ordinary differential equations ( [ theeqn ] )
. it also has an adjoint : @xmath111 note that while the orthogonality in eq .
( [ invmeasure ] ) indicates that @xmath112 is a stationary solution to eq .
( [ thepde2 ] ) , it is not a stationary invariant density to ( [ thepde ] ) .
this is due to the fact that vector field @xmath113 is not divergence free , but rather as in ( [ non - df ] ) the scalar factor @xmath30 .
then it is easy to verify that @xmath114 is a stationary solution to ( [ thepde ] ) : @xmath115 it is widely known that a volume - preserving , divergence - free conservative dynamics has a conserved entropy @xmath116 = -\int_{\mathbb{r } } u(x , t)\ln u(x , t)\rd x$ ] @xcite . for conservative system like ( [ theeqn ] ) which contains the scalar factor @xmath117 , the shannon entropy should be replaced by the relative entropy with respect to @xmath35 ( see appendix [ sec6.2 ] for detailed calculation ) : @xmath118 such systems are called _ canonical conservative _ with respect to @xmath35 in @xcite . in classical statistical physics , the term @xmath119 is called free energy @xcite ; in information theory , kullback - leibler divergence .
we can in fact show a stronger result , with arbitrary differentiable @xmath120 and @xmath121 over an arbitrary domain @xmath122 ( see appendix [ sec6.2 ] ) : @xmath123 & = & \int_{\partial\mathfrak{d } } \left\ { u(x , y , t ) \psi\left ( \frac{u(x , y , t)}{g^{-1}(x , y)\rho(h)}\right ) \big(f , g\big ) \right\ } \times ( \rd x,\rd y ) .
\label{eq34}\end{aligned}\ ] ] therefore , if @xmath124 , then @xmath125 , and the integral on the right - hand - side of ( [ eq34 ] ) is always zero . in other words , in conservative dynamics like ( [ theeqn ] ) , it is the support @xmath126 on which @xmath127 is observed that determines whether a system is invariant ; not the initial data @xmath128 @xcite . since a `` state '' is defined as an entire orbit , it is natural to change the coordinates from @xmath43 to @xmath129 according to the solution curve to ( [ theeqn ] ) , where we use @xmath130 to denote time , @xmath131 .
we have @xmath132 @xmath133 therefore : @xmath134 = xy = g(x , y ) .
\label{det}\ ] ] then , the generalized relative entropy can be expressed as @xmath135 in which @xmath136 and @xmath137 @xmath138 is known as boltzmann s entropy in classical statistical mechanics .
we see that the @xmath2 introduced in sec .
[ sec2 ] is the simplest case of the generalized relative entropy in ( [ eq34 ] ) with @xmath139 , and @xmath140 .
then @xmath141 .
gibbs canonical ensemble chooses @xmath142 .
the dynamics ( [ theeqn ] ) is not ergodic in the @xmath7-plane ; it does not have a unique invariant measure , as indicated by the arbitrary @xmath37 in eq .
( [ non - ergodic ] ) . however , the function @xmath117 , as indicated in eqs .
( [ t2that ] ) and ( [ det ] ) , is the unique invariant measure on each ergodic invariant submanifold @xmath17 .
it is non - uniform with respect to lebesgue measure . on the ergodic invariant manifold @xmath17 : @xmath143 . to see the difference between the lebesgue - based average and invariant - measure based average , consider a simple time - varying exponentially growing population : @xmath144 .
the regular time average of the per capita growth rate is @xmath145 the lebesgue - based average is an `` average growth rate per average capita '' @xmath146 in cyclic population dynamics , this latter quantity corresponds to the @xmath147 weighted per capita growth rate or kinetic energy " @xmath148 = & & \frac{\displaystyle\int_0^{\tau(h,\alpha ) } x\left(\frac{\partial h}{\partial x}\right ) g\big(x(t),y(t)\big ) \rd t } { \displaystyle\int_0^{\tau(h,\alpha ) } g\big(x(t),y(t)\big ) \rd t } = \frac{\displaystyle\int_0^{\tau(h,\alpha ) } y \left(\frac{\partial h}{\partial y } \right ) g\big(x(t),y(t)\big ) \rd t } { \displaystyle\int_0^{\tau(h,\alpha ) } g\big(x(t),y(t)\big ) \rd t}.\end{aligned}\ ] ]
in this section , we show that the conservative dynamics in ( [ theeqn ] ) is an emergent caricature of a robust stochastic population dynamics . this material can be found in many texts , e.g. , @xcite .
but for completeness , we shall give a brief summary .
assume the populations of the prey and the predator , @xmath149 and @xmath150 , reside in a spatial region of size @xmath151 .
the discrete stochastic population dynamics follows a two - dimensional , continuous time birth - death process with transition probability rate @xmath152 & = & \left ( m\delta_{k , m+1 } + \dfrac{1}{\omega } mn \delta_{k , m-1 } + \dfrac{\alpha}{\omega } nm \delta_{\ell , n+1 } + \alpha n\delta_{\ell , n-1 } \right)\delta t + o(\delta t ) .
\label{bdp}\end{aligned}\ ] ] the discrete stochastic dynamics has an invariant measure : @xmath153 then @xmath154 \nonumber\\[6pt ] & + & p_{m-1,n}(t)\big[(m-1)\delta t\big ] + p_{m+1,n}(t ) \left[\dfrac{1}{\omega}(m+1)n\delta t\right ] \nonumber\\[6pt ] & + & p_{m , n-1}(t)\left[\dfrac{\alpha}{\omega}(n-1)m\delta t\right ] + p_{m , n+1}(t)\big[\alpha(n+1)\delta t\big ] .
\nonumber\end{aligned}\ ] ] that is @xmath155 -p_{m-1,n}(t ) \nonumber\\[6pt ] & + & \dfrac{1}{\omega } mn \big[p_{m+1,n}(t)-p_{m , n}(t ) \big ] + \dfrac{1}{\omega } n p_{m+1,n}(t ) \nonumber\\[6pt ] & - & \dfrac{\alpha}{\omega } nm \big[p_{m , n}(t)-p_{m , n-1}(t ) \big ] - \dfrac{\alpha}{\omega } m p_{m , n-1}(t ) \nonumber\\[6pt ] & + & \alpha n \big[p_{m , n+1}(t)-p_{m , n}(t ) \big ] + \alpha p_{m , n+1}(t ) .
\label{eq49}\end{aligned}\ ] ] for a very large @xmath151 , the population _ densities _ at time @xmath156 can be approximated by continuous random variables as @xmath157 and @xmath158 .
( [ eq49 ] ) becomes a partial differential equation by setting @xmath159 , @xmath160 , and @xmath161 : @xmath162 & + & xy\dfrac{\partial u}{\partial x } + \dfrac12 \omega^{-1 } xy \dfrac{\partial^2 u}{\partial x^2 } + y u + \omega^{-1 } y \dfrac{\partial u}{\partial x } \nonumber\\[6pt ] & - & \alpha xy \dfrac{\partial u}{\partial y } + \dfrac\alpha2 \omega^{-1 } xy \dfrac{\partial^2 u}{\partial y^2 } - \alpha x u + \alpha \omega^{-1 } x \dfrac{\partial u}{\partial y } \nonumber\\[6pt ] & + & \alpha y \dfrac{\partial u}{\partial y } + \dfrac\alpha2 \omega^{-1}y \dfrac{\partial^2 u}{\partial y^2 } + \alpha u + \alpha \omega^{-1}\dfrac{\partial u}{\partial y } + o(\omega^{-1 } ) .
\nonumber\end{aligned}\ ] ] rearranging the terms and writing @xmath163 , we can perform the kramers - moyal expansion to obtain : @xmath164 with drift @xmath165 and symmetric diffusion matrix @xmath166 eq .
( [ thefpe ] ) should be interpreted as a fokker - plank equation for the probability density function @xmath167 .
it represents a continuous stochastic process @xmath168 following it integral @xcite : @xmath169 \label{sde } \\[-7pt ] \rd y(t ) & = & \alpha y(x-1 ) \rd t + \epsilon^{\frac{1}{2}}\sqrt{\alpha y(x+1)}\ \rd w_2(t ) \nonumber\end{aligned}\ ] ] it is important to recognize that in the limit of @xmath170 , the dynamics described by eq .
( [ thefpe ] ) is reduced to that in eq .
( [ thepde ] ) , which is equivalent to eq .
( [ theeqn ] ) via the method of characteristics .
it can be verified that the stationary solution to ( [ thefpe ] ) is actually @xmath171 , which is consistent with the discrete case ( cf .
. [ bdp ] ) , and also a stationary solution to the liouville equation eq .
( [ thepde ] ) . as suggested in @xcite , the right - hand - side of eq .
( [ thefpe ] ) has a natural decomposition : @xmath172 \nonumber\\ & = & \nabla\cdot \big [ \epsilon\md \big(u\nabla\ln u + u \nabla\ln g \big ) - \mf u \big ] \nonumber\\ & = & \epsilon\nabla\cdot \md u \nabla \big(\ln \big ( g \ u \big ) \big ) - \nabla\cdot \big ( \mf \ u \big)\end{aligned}\ ] ] in which the first term is a self - adjoint differential operator and the second is skew - symmetric @xcite .
the equation from the first line to the second uses the fact @xmath173 , thus @xmath174 . in terms of the stochastic differential equation in divergence form
, this decomposition corresponds to : @xmath175 \rd y \end{array}\right ) = -\epsilon\md \nabla \ln g + g \left(\begin{array}{c } -h_y \\[5pt ] h_x \end{array}\right ) + \epsilon^{\frac{1}{2 } } \sqrt{2\md } \left(\begin{array}{c } \rd w_1(t ) \\[5pt ] \rd w_2(t ) \end{array}\right ) .
\label{newsde}\ ] ] under this non - it interpretation of the stochastic differential equation , the finite population with fluctuations ( i.e. , @xmath176 ) is unstable when @xmath177 .
the system behaves as an unstable focus as shown in fig .
[ fig_3 ] .
the eigenvalues at the fixed point @xmath178 are @xmath179 , corresponding to the unstable nature of the stochastic system . ) , the deterministic part of system ( [ newsde ] ) .
red curves denote phase flow and blue arrows denote the strength of the vector field.,width=240 ] on the other hand , the potential - current decomposition reveals that the system ( [ theeqn ] ) will be structurally stable in terms of the stochastic model : any perturbation of the model system will yield corresponding conserved dynamics close to ( [ theeqn ] ) .
the conservative ecology is a robust emergent phenomenon .
equations such as ( [ sde ] ) and ( [ newsde ] ) are not mathematically well - defined until an precise meaning of integration @xmath180 is prescribed .
this yields different stochastic processes @xmath181 whose corresponding probability density function @xmath182 follow different linear partial differential equations .
the fundamental solution to any partial differential equation ( pde ) , however , provides a markov transition probability ; there is no ambiguity at the pde level . on the other hand , the only interpretation of ( [ wt ] ) that provides a markovian stochastic process that is non - anticipating is that of it s @xcite .
the differences in the interpretations of ( [ wt ] ) become significant only in the modeling context , when one s intuition expects that @xmath183=0 $ ] even for interpretations other then it s . with the @xmath168 defined in ( [ thefpe ] ) and ( [ sde ] ) ,
let us now consider the stochastic functional @xmath184 therefore , for very large populations , i.e. , small @xmath185 , this suggests a separation of time scales between the cyclic motion on @xmath17 and slow , stochastic level crossing @xmath186 .
the method of averaging is applicable here @xcite : @xmath187 with @xmath188 where @xmath189 denotes the average of @xmath58 on the level set @xmath17 .
then , using the it integral , the distribution of @xmath186 follows a fokker - planck equation : @xmath190 and the stationary solution for eq .
( [ varyingh ] ) is : @xmath191 the steady state distributions of @xmath27 under different @xmath0 s are shown in fig .
[ fig_3 ] .
the steady state distribution @xmath192 does not depend on the volume size @xmath193 .
, the steady state distribution @xmath192 with respect to @xmath27 in logarithmic scale .
the slowly fluctuating energy " @xmath27 ranges from @xmath194 to infinity .
its steady state distribution @xmath192 eventually increases without bound as @xmath27 increases.,width=336 ] when @xmath27 is big enough , @xmath192 increases with @xmath27 without bound , since @xmath195 is a positive increasing function .
hence , @xmath192 is not normalizable on the entire @xmath196 , reflecting the unstable nature of the system .
the fluctuation @xmath197 approaches zero when @xmath27 approaches @xmath194 .
consequently , the absorbing effect at @xmath198 makes @xmath199 another possible local maximum .
it is usually an obligatory step in understanding an ode @xmath200 to analyze the dependence of a steay state @xmath201 as an implicit function of the parameters @xmath202 @xcite .
one of the important phenomena in this regard is the thom - zeeman catastrophe @xcite . from this broad perspective
, the analysis developed by helmholtz and boltzmann in 1884 is an analysis of the geometry of a `` non - constant but steay solution '' , as a function of its parameter(s ) and initial conditions . in the context of lv equation ( [ theeqn ] ) ,
the geometry is characterized by the area encircled by a periodic solution , @xmath17 , where @xmath75 is specified by the the initial data : @xmath203 .
the celebrated helmholtz theorem @xcite then becomes our eq .
( [ dheq28 ] ) @xmath204 since eq .
( [ theeqn ] ) has a conserved quantity @xmath27 , eq . ( [ dheq ] ) can , and should be , interpreted as an extended @xmath27 conservation law , beyond the dynamics along a single trajectory , that includes both variations in @xmath0 and in @xmath75 . the partial derivatives in ( [ dheq ] )
can be shown as time averages of ecological activeness @xmath205 or @xmath206 , and variation in the prey s population @xmath207 .
those conjugate variables , along with parameter @xmath0 , conserved quantity @xmath27 , and encompassed area @xmath208 constitutes a set of state variables " describing the state of an ecological system in its stationary , cyclic state .
this is one of the essences of boltzmann s statistical mechanics @xcite . for the monocyclic lotka - volterra system ,
the dynamics are relatively simple .
hence , the state variables have monotonic relationships , the same as that observed in ideal gas models .
when the system s dynamics become more complex ( e.g. have more than one attractor , hopf bifurcation ) , relations among the state variables will reflect that complexity ( e.g. develop a cusp , exhibiting a phase transition in accordance @xcite ) . when the populations of predator and prey are finite , the stochastic predator - prey dynamics is unstable . this fact
is reflected in the non - normalizable steady state distribution @xmath35 on @xmath209 , and the destabilizing effect of the gradient dynamics in the potential - current decomposition .
this is particular to the lv model we use ; it is not a problem for the general theory if we study a more realistic model as in @xcite . despite the unstable dynamics ,
the stochastic model system is structurally stable : its dynamics persists under sufficiently small perturbations .
this implies conservative dynamical systems like ( [ theeqn ] ) are meaningful mathematical models , when interpreted correctly , for ecological realities .
indeed , all ecological population dynamics can be represented by birth - death stochastic processes @xcite . except for systems with detailed balance , which rarely holds true , almost all such dynamics have underlying cyclic , stationary conservative dynamics .
the present work shows that a hidden conservative ecological dynamics can be revealed through mathematical analyses .
to recognize such a conservative ecology , however , several novel quantities need to be defined , developed , and becoming a part of ecological vocabulary .
this is the intellectual legacy of helmholtz s and boltzmann s mechanical theory of heat @xcite
. 0.3 cm * authors contributions . *
m. and h. q. contributed equally to this work .
* competing interests . *
we declare there are no competing interests
. * acknowledgements . *
we thank r.e .
omalley , jr . and l.f .
thompson for carefully reading the manuscripts and many useful comments .
99 murray , j. d. ( 2002 ) _ mathematical biology i : an introduction _ , 3rd ed . , springer , new york .
kot , m. ( 2001 ) _ elements of mathematical ecology _ , cambridge univ .
press , uk .
qian , h. ( 2011 ) nonlinear stochastic dynamics of mesoscopic homogeneous biochemical reaction systems - an analytical theory ( invited article ) .
_ nonlinearity _ , * 24 * , r19r49 .
( doi:10.1088/0951 - 7715/24/6/r01 ) zhang , x .-
j . , qian , h. and qian , m. ( 2013 ) stochastic theory of nonequilibrium steady states and its applications .
part i. _ phys .
rep . _ * 510 * , 185 .
( doi:10.1016/j.physrep.2011.09.002 ) ao , p. ( 2005 ) laws in darwinian evolutionary theory ( review ) . _ phys .
_ * 2 * , 117156 .
( doi:10.1016/j.plrev.2005.03.002 ) qian , h. , qian , m. and wang , j .- z .
( 2012 ) circular stochastic fluctuations in sis epidemics with heterogeneous contacts among sub - populations .
_ theoret .
biol . _ * 81 * , 223231 .
( doi:10.1016/j.tpb.2012.01.002 ) wang , j. , xu , l. and wang , e. ( 2008 ) potential landscape and flux framework of nonequilibrium networks : robustness , dissipation , and coherence of biochemical oscillations .
usa _ * 105 * , 1227112276 .
( doi:10.1073/pnas.0800579105 ) qian , h. ( 2013 ) a decomposition of irreversible diffusion processes without detailed balance . _ j. math
_ * 54 * , 053302 .
( doi:10.1063/1.4803847 ) qian , h. ( 2014 ) the zeroth law of thermodynamics and volume - preserving conservative system in equilibrium with stochastic damping .
a _ * 378 * , 609616 . ( doi:10.1016/j.physleta.2013.12.028 ) gallavotti , g. ( 1999 ) _ statistical mechanics : a short treatise _ , springer , berlin .
khinchin , a. i. ( 1949 ) _ mathematical foundations of statistical mechanics _ , dover , new york .
campisi , m. ( 2005 ) on the mechanical foundations of thermodynamics : the generalized helmholtz theorem . _ studies in history and philosophy of modern physics _ , * 36 * , 275290 .
( doi:10.1016/j.shpsb.2005.01.001 ) ma , y .- a . and qian , h. ( 2015 ) universal ideal behavior and macroscopic work relation of linear irreversible stochastic thermodynamics .
_ new j. phys . _ * 17 * , 065013 .
( doi:10.1088/1367 - 2630/17/6/065013 ) lotka , a. j. ( 1925 ) _ elements of physical biology _ , williams & wilkins , baltimore , md .
planck , m. ( 1969 ) _ treatise on thermodynamics _
, dover , new york .
pauli , w. ( 1973 ) _ pauli lectures on physics : thermodynamics and the kinetic theory of gas _ , vol .
3 , the mit press , cambridge , ma . allen , l. j. s. ( 2010 ) _ an introduction to stochastic processes with applications to biology _
, chapman & hall / crc , new york .
grasman , j. and van herwaarden , o. a. ( 1999 ) _ asymptotic methods for the fokker - planck equation and the exit problem in applications _ , springer , berlin .
kurtz , t. g. ( 1981 ) _ approximation of population processes .
_ siam pub , philadelphia .
qian , h. ( 2015 ) thermodynamics of the general diffusion process : equilibrium supercurrent and nonequilibrium driven circulation with dissipation .
j. special topics _ , * 224 * , 781799 .
( doi:10.1140/epjst / e2015 - 02427 - 6 ) andrey , l. ( 1985 ) the rate of entropy change in non - hamiltonian systems . _
lett . a _ * 111 * , 4546 .
( doi:10.1016/0375 - 9601(85)90799 - 6 ) zhang , f. , xu , l. , zhang , k. , wang , e. , and wang , j. ( 2012 ) the potential and flux landscape theory of evolution .
_ j chem phys . _ * 137 * , 065102 .
( doi:10.1063/1.4734305 ) gardiner , c. w. ( 2004 ) _ handbook of stochastic methods : for physics , chemistry and the natural sciences _ , 3rd ed . , springer - verlag , berlin .
freidlin , m. i. and wentzell , a. d. ( 2012 ) _ random perturbations of dynamical systems _
, springer - verlag , new york .
zhu , w.q .
( 2006 ) nonlinear stochastic dynamics and control in hamiltonian formulation . _
_ , * 59 * , 230248 .
ao , p. , qian , h. , tu , y. , and wang , j. ( 2013 ) a theory of mesoscopic phenomena : time scales , emergent unpredictability , symmetry breaking and dynamics across different levels , arxiv:1310.5585 .
xu , l. , zhang , f. , zhang , k. , wang , e. , and wang , j. ( 2014 ) the potential and flux landscape theory of ecology .
_ plos one _ * 9 * , e86746 .
( doi:10.1371/journal.pone.0086746 ) gallavotti , g. , reiter , w. l. and yngvason , j. ( 2007 ) _
boltzmann s legacy _ ,
esi lectures in mathematics and physics , eur .
soc . pub .
[ sec : aa ] @xmath210 similarly @xmath211
[ sec6.2 ] using the divergence theorem and noting that @xmath212 , we obtain for the time evolution of the relative entropy : @xmath213 & = & \int_{\mathbb{r}^2}\frac{\partial u(x , y , t)}{\partial t } \left[\ln\left ( \frac{u(x , y , t)}{g^{-1}(x , y)}\right ) + 1 \right ] \rd x\rd y \nonumber\\[6pt ] & = & -\int_{\mathbb{r}^2 } \nabla\cdot\big ( ( f , g)u(x , y , t)\big ) \ln\left ( \frac{u(x , y , t)}{g^{-1}(x , y)}\right ) \rd x\rd y + \frac{\partial}{\partial t}\int_{\mathbb{r}^2 } u(x , y , t ) \rd x\rd y \nonumber\\[6pt ] & = & -\int_{\mathbb{r}^2 } \nabla\cdot\big ( ( f , g)u(x , y , t)\big ) \ln\left ( \frac{u(x , y , t)}{g^{-1}(x , y)}\right ) \rd x\rd y\ \nonumber\\[6pt ] & = & -\int_{\mathbb{r}^2 } \nabla\cdot\big(g^{-1}f , g^{-1}g\big ) \times \left(\frac{u}{g^{-1}}\right ) \ln\left(\frac{u}{g^{-1}}\right ) \rd x\rd y \nonumber\\[6pt ] & & -\int_{\mathbb{r}^2 } \big(g^{-1}f , g^{-1}g\big ) \cdot \nabla\left[\left(\frac{u}{g^{-1}}\right ) \ln\left(\frac{u}{g^{-1}}\right ) - \left(\frac{u}{g^{-1}}\right)\right ] \rd x\rd y \nonumber\\[6pt ] & = & \int_{\mathbb{r}^2 } \left [ \left(\frac{u}{g^{-1}}\right ) \ln\left(\frac{u}{g^{-1}}\right)-\left(\frac{u}{g^{-1}}\right ) \right ] \nabla\cdot \big(g^{-1}f , g^{-1}g\big ) \
\rd x\rd y \
= \ 0.\end{aligned}\ ] ] a more general result can be obtained in parallel for arbitrary differentiable @xmath214 and @xmath215 : @xmath123 & = & \int_{\mathfrak{d } } \frac{\partial u(x , y , t)}{\partial t } \left [ \psi\left ( \frac{u}{g^{-1}\rho(h)}\right ) + \frac{u}{g^{-1}\rho(h ) } \psi'\left ( \frac{u}{g^{-1}\rho(h)}\right ) \right ] \rd x\rd y \nonumber\\[6pt ] & = & -\int_{\mathfrak{d } } \nabla\cdot\big ( ( f , g ) u\big ) \left [ \psi\left ( \frac{u}{g^{-1}\rho(h)}\right ) + \frac{u}{g^{-1}\rho(h ) } \psi'\left ( \frac{u}{g^{-1}\rho(h)}\right ) \right ] \rd x\rd y \nonumber\\[6pt ] & = & -\int_{\mathfrak{d } } \big(g^{-1}\rho(h)f , g^{-1}\rho(h)g\big ) \cdot \nabla\left(\frac{u}{g^{-1}\rho(h ) } \right ) \nonumber\\ & & \times \left [ \psi\left ( \frac{u}{g^{-1}\rho(h)}\right ) + \frac{u}{g^{-1}\rho(h ) } \psi'\left ( \frac{u}{g^{-1}}\rho(h)\right ) \right ] \rd x\rd y \nonumber\\[6pt ] & = & -\int_{\mathfrak{d } } \big(g^{-1}\rho(h)f , g^{-1}\rho(h)g\big ) \cdot \nabla \left [ \frac{u}{g^{-1}\rho(h ) } \psi\left ( \frac{u}{g^{-1}\rho(h)}\right ) \right ] \rd x\rd y \nonumber\\[6pt ] & = & -\int_{\mathfrak{d } } \nabla\cdot\left\ { \big ( g^{-1}\rho(h)f , g^{-1}\rho(h)g\big ) \left [ \frac{u}{g^{-1}\rho(h ) } \psi\left ( \frac{u}{g^{-1}\rho(h)}\right ) \right ] \right\ } \rd x\rd y \nonumber\\[6pt ] & = & \int_{\partial\mathfrak{d } } \left\ { u(x , y , t ) \psi\left ( \frac{u(x , y , t)}{g^{-1}(x , y)\rho(h)}\right ) \big(f , g\big ) \right\ } \times ( \rd x,\rd y).\end{aligned}\ ] ] | we carry out mathematical analyses , _ la _ helmholtz s and boltzmann s 1884 studies of monocyclic newtonian dynamics , for the lotka - volterra ( lv ) equation exhibiting predator - prey oscillations . in doing so a novel thermodynamic theory " of ecology
is introduced .
an important feature , absent in the classical mechanics , of ecological systems is a natural stochastic population dynamic formulation of which the deterministic equation ( e.g. , the lv equation studied ) is the infinite population limit .
invariant density for the stochastic dynamics plays a central role in the deterministic lv dynamics .
we show how the conservation law along a single trajectory extends to incorporate both variations in a model parameter @xmath0 and in initial conditions : helmholtz s theorem establishes a broadly valid conservation law in a class of ecological dynamics .
we analyze the relationships among mean ecological activeness @xmath1 , quantities characterizing dynamic ranges of populations @xmath2 and @xmath0 , and the ecological force @xmath3 .
the analyses identify an entire orbit as a stationary ecology , and establish the notion of equation of ecological states " .
studies of the stochastic dynamics with finite populations show the lv equation as the robust , fast cyclic underlying behavior .
the mathematical narrative provides a novel way of capturing long - term dynamical behaviors with an emergent _ conservative ecology_. = 6 . in = 9.0 in = -0.5 in = -0.85 in = 0.23 in * key words : * conservation law , ecology , equation of states , invariant density , stochastic thermodynamics |
in the mark iii catalog ( willick et al . 1995 , 1996 ) the distance of the virgo cluster ( vc ) is @xmath1 km s@xmath0 .
the heliocentric radial velocity of the vc ( @xmath2 ) from huchra ( 1996 ) is @xmath3 km s@xmath0 .
combine this with the heliocentric peculiar motion of the sun in the cosmic microwave background ( cmb ) frame of 368.6 km s@xmath0 towards l = 264.7@xmath4 , b = 48.2@xmath4 , yields a cmb radial velocity for the vc ( @xmath5 ) of @xmath6 km s@xmath0 .
if we combine @xmath5 with the peculiar velocity of the local group ( lg ) in the cmb frame ( 372 km s@xmath0 ) in the direction of the vc , we find the lg and the vc move towards each other in the cmb reference frame with a mutual velocity of @xmath7 km s@xmath0 .
four issues can be addressed from these data : \1 ) how much of the cmb motion between the lg and the vc is generated by the vc itself ( i.e. , classic `` virgocentric infall '' motion ) , how much is generated by masses on larger size scales ?
\2 ) the peculiar velocities of the galaxy groups defined to be within the coma - sculptor ( c - s ) cloud by tully ( 1987 ) can be demonstrably shown to share most of the lg cmb peculiar velocity ( e.g. faber & burstein 1988 ; tully 1988 ) .
much of this motion ( @xmath8 km s@xmath0 ) is directed perpendicular to the supergalactic plane .
\3 ) as viewed in the lg reference frame , galaxy groups within the c - s cloud are moving towards each other .
how much of these motions are generated by masses within the c - s cloud and how much are externally generated ?
\4 ) in the direction towards the ursa major ( umaj ) cluster ( + x , + y sg coordinates ) , all galaxies with @xmath9 km s@xmath0 have peculiar velocities moving towards the lg .
how much of these inward motions are generated by mass within the local supercluster ( lsc ) , and how much by mass outside it ?
why rexamine these issues now ? based on the mark iii database and on a new potent methodology , dekel et al .
( 1999 ) have generated a more accurate model of the peculiar motions generated _ within _ the lsc by masses _
outside _ the lsc . for the present paper , ami eldar and avishai dekel
have kindly calculated from this model the `` outside region(generated)motions '' ( o - r - m ) for galaxies within a distance of 3000 km s@xmath0 from the masses outside this region . for this analysis ,
we use only the spiral galaxies in the mark iii database .
we use the cmb velocity reference frame only for galaxies outside the c - s cloud .
the lg velocity velocity reference frame is used for galaxies inside the c - s cloud ( using the yahil , tammann & sandage ( 1977 ) vector for the heliocentric to lg transformation ) , as these galaxies demonstrably share the lg peculiar velocities .
the o - r - m peculiar velocity of the lg is , in the cmb frame , ( -399.5 , 139.3 , -52.4 ) km s@xmath0 in ( x , y , z ) supergalactic coordinates , or 426.3 km s@xmath0 towards l = 160.8@xmath4 , b = -7.0@xmath4 ( galactic l = 307.8@xmath4 , b = 18.2@xmath4 ) . while this is essentially the same direction as found by lynden - bell et al .
( 1988 ) in the original 7samurai analysis , the overall o - r - m velocity field predicts a lower peculiar velocity for the lg in this direction , as well as a more complex velocity field outside and within the lsc than the great attractor model used there and by faber & burstein ( 1988 ) . in the o - r - m model
, the vc has a net cmb motion relative to the lg of -178 km s@xmath0 , resulting in a vc peculiar velocity _ in the direction of _
( i.e. , not relative to ) the lg of @xmath10 km s@xmath0 .
the o - r - m model also gives the lg a peculiar velocity towards the direction of the vc of 227 km s@xmath0 .
this leaves a net peculiar velocity of the lg _ in the direction of _ the vc of @xmath11 km s@xmath0 .
note that , if we insist on keeping the vc peculiar motion to be zero in the cmb / o - r - m frame of motion , the net motion _ between _ the lg and the vc is 145 km s@xmath0 .
however , if we take at face value the calculated peculiar motion of the vc relative to the cmb / o - r - m frame ( -88 km s@xmath0 ) , we get the net motion _ between _ the lg and the vc to be @xmath12 km s@xmath0 . in other words , if we remove the outside gravitational influences from the virgocentric infall issue , the local group is moving towards the virgo cluster at a speed of @xmath13 km s@xmath0 , and the vc itself is moving 88 km s@xmath0 towards the lg . to model the residual velocity field within the lsc relative to the o - r - m model we use a virgocentric infall ( vinfall ) model based on that used by lynden - bell et al .
( a linearized model of that proposed by schechter 1980 ) . in this model , we nominally assign a lg virgocentric infall velocity of 220 km s@xmath0 , leaving a residual of @xmath14 km s@xmath0 of lg motion relative to the vc ; i.e. , well within the noise .
likewise , the combined o - r - m+vinfall model leaves a net lg motion towards the center of umaj of + 15 km s@xmath0 , with an error at least as large as that towards the vc . as one will see in the graphs at the end of the paper , the combined o - r - m + vinfall model accounts for all but @xmath15 km s@xmath0 systematic motions _ within _ the lsc , also within the known noise of galaxy peculiar motions
the cmb motion of the lg is 592 km s@xmath0 towards l = 276.0@xmath4 , b = 23.6@xmath4 ( l = 146.0@xmath4 , b = -33.9@xmath4 ) . subtracting the lg o - r - m motion , 426 km s@xmath0 towards l = 307.8@xmath4 , b = 18.2@xmath4 ( l = 160.8@xmath4 , b = -7.0@xmath4 )
, we get a net motion of the lg relative to the o - r - m model of 309 km s@xmath0 towards l = 228.9@xmath4 , b = 19.8@xmath4 ( l = 93.5@xmath4 , b = -63.9@xmath4 ) . removing a virgocentric infall of 220 km s@xmath0 towards l = 283.8@xmath4 ,
b = 74.5@xmath4 ( l = 102.9@xmath4 , b = -2.4@xmath4 ) yields a net motion of the lg relative to the combined o - r - m+vinfall models of 273 km s@xmath0 towards l = 222.9@xmath4 , b = -4.7@xmath4 ( l = 18.9@xmath4 , b = -85.3@xmath4 ) . because both the o - r - m vector and the vinfall vectors lie so close to the supergalactic plane ( 7@xmath4 and 2.4@xmath4 , respectively ) , the -330.1 km s@xmath0 z - component of the lg cmb motion can not be fit by either component .
thus , we are left with a component of the local group s cmb velocity that is larger than its virgocentric infall velocity , in a direction almost exactly perpendicular to the supergalactic plane - the local anomaly . the problem with understanding
this part of the local anomaly motion lies , of course , with the direction it is located within our galaxy - within 5@xmath4 of the galactic plane .
given that the supergalactic plane is nearly perpendicular to the galactic plane , it comes as little suprise that almost all of the galaxies known within the lsc lie within the supergalactic plane .
here we concentrate on presenting the residual motions within the lsc _ after _ this model is subtracted from the observed peculiar velocity field .
the interested reader is referred to burstein et al .
( 1996 ) for how the peculiar velocity field appears before this model - fit . as in burstein et al . , we restrict our picture of the peculiar of motions of lsc galaxies to those galaxies lying within @xmath16 of the supergalactic plane , and within @xmath17 of the supergalactic y direction .
the planar cut removes issues pertaining to galaxy infall relative to the supergalactic plane ; the y - direction cut restricts our view to those motions that are primarily directed towards the two largest masses within the lsc vc and umaj .
figure 1 shows the peculiar velocity field relative to the o - r - m+vinfall model , as viewed from the direction of the north supergalactic pole , with peculiar velocities of the galaxies plotted in their full ( i.e. , not _ projected _ velocities ) , a la@xmath18 lynden - bell et al . for this view alone
, we include the peculiar velocities of the galaxies known to be in the vc .
figure 2a shows the peculiar velocity field for the same galaxies as in figure 1 , relative to just the o - r - m model .
2a views the peculiar velocities along the y - axis as if removed from the flow , negating the sign of peculiar velocities in the -y direction .
comparison of this figure with fig .
5a in burstein et al . shows that the o - r - m model removes the apparent infall towards the lg for galaxies with @xmath19 km s@xmath0 .
the o - r - m model now makes it clear that the apparent infall velocities seen along the y axis are primarily a lsc phenomenon , rather than one of a larger region .
figure 2b shows the peculiar velocity field along the y axis relative to the combined o - r - m+vinfall model velocity field ( in this and subsequent views , galaxies are excluded if they lie within 5.7@xmath4 of the vc direction , and within 530 km s@xmath0 of the vc distance ) .
as stated earlier , it is apparent that the combination of the o - r - m + vinfall velocity fields fit well the velocity field within the lsc , modulo residual motions of @xmath20 km s@xmath0 remaining , as evident by the coherent residual motions of the galaxies within the coma - sculptor cloud .
figures 3a d show the views in a 35@xmath4 cone towards the vc ( left - hand side ) and in a 25@xmath4 cone towards umaj ( rhs ) .
the top graphs are the peculiar velocities of galaxies as seen in the observed cmb + local anomaly reference frame .
the bottom graphs are the peculiar velocities of the galaxies in these directions relative to the combined o - r - m+vinfall model .
it is clear the combined model not only removes all evidence of infall on the near side of the vc , it also removes essentially all of the inwards motions of galaxies in the the umaj direction .
there is even a suggestion in fig .
3d of a slight infall pattern relative to umaj itself .
the well - known quadrupole peculiar velocity field within the lsc is well - fit by a a combination of two physically - generated motions : a ) those generated by mass distributed around the local supercluster at @xmath21 km s@xmath0 ( from the o - r - m model ) ; and b ) the virgocentric infall ( vinfall ) pattern from a virgo cluster moving at 88 km @xmath0 motion towards the lg , which produces 220 km s@xmath0 of infall of the lg towards the vc .
( if we insist the vc have no peculiar motion in the o - r - m model , there would be an lg infall velocity to the vc in the o - r - m model of 132 km s@xmath0 . )
the strong observed infall motion of the ursa major region is well - fit by the combined o - r - m+vinfall model _ plus _ the net infall motion of the virgo cluster towards the lg .
if one insists on putting the virgo cluster at rest relative to the model , all the galaxies in the ursa major region are then given an overall 88 km s@xmath0 peculiar motion towards the lg .
two kinds of coherent motions remain relative to the combined model : a ) @xmath15 km s@xmath0 motions , coherent over 1000 - 1500 km s@xmath0 size scales ( e.g. , what we see for the c - s cloud ) .
b ) the local anomaly , now seen as a 270 km s@xmath0 motion of the local group nearly perpendicular to the supergalactic plane . is this sgz motion just that of the local group , does it involve the whole coma - sculptor cloud , or does it involved the whole lsc ? while the `` local void '' ( cf .
tonry & dressler , this meeting ) is likely a contributant to the local anomaly motion , for it to `` push '' the lg stronger than the vc `` pulls '' us seems unlikely . as the bulk of our knowledge of lsc motions lie within the supergalactic plane , it is improbable for us to know the full origin of the local anomaly without knowing the distances of galaxies behind the zone of avoidance of the galactic plane .
saito et al . ( 1991 ) and lahav et al . (
1993 ) have shown that near l = 230@xmath4 , b = -4@xmath4 is a galaxy cluster in puppis , heavily extincted by the galaxy and masked by foreground stars .
given this cluster appears strongly in iras - detected galaxies , it is likely that it also contains a substantial number of early - type galaxies .
if we can measure the distances to galaxies in the puppis region , we can test whether the sgz motion of the lg is shared by the whole lsc or not .
if it is just the lg / c - s cloud doing this motion , then we should simply see the reflex motion of the sgz reflected in the peculiar velocities of the puppis galaxies .
if , on the other hand , the lsc and the puppis supercluster are being gravitationally drawn together , the galaxies in puppis will show peculiar velocities towards us in substantial excess of the cmb reflex .
this is a test definitely worth doing , albeit the known difficulties involved .
this work could not have been done withont the efforts of the whole mark iii team ( jeff willick , sandy faber , stephane courteau , avishai dekel , michael strauss , tsafrir kolatt and bepi tormen ) , and especially the o - r - m calculations provided to the author by ami eldar and avishai dekel .
my heartfelt thanks to all of them .
burstein d , et al .
1996 , in gravitational dynamics , eds .
o. lahav , e. terlevich & r.j .
terlevich , ( cambridge university press : cambridge ) , p. 141 dekel , a. , eldar , a. , kolatt , t. , willick , j.a . ,
faber , s.m . , courteau , s. , burstein , d. 1999 , apj 522 , 000 faber , s.m .
& burstein , d. 1988 , in large - scale motions in the universe , eds .
v.c . rubin and g.v .
coyne , ( princeton univ : princeton ) , p. 115 huchra , j.p .
1996 , in peculiar velocities in the universe , heron island workshop , in press lahav , o. , yamada , t. , scharf , c. , & kraan - korteweg , r.c .
1993 , mnras 262 , 711 lynden - bell , d. , faber , s.m . ,
burstein , d. , davies , r.l . , dressler , a. , terlevich , r. & wegner , g. 1987 , apj 326 , 19 saito , m. et al .
1991 , pasj 43 , 449 schechter , p.l 1980 , aj 85 , 801 tully , r.b .
1987 , nearby galaxies catalog ( cambridge university press : new york ) tully , r.b .
1988 , in large - scale motions in the universe , eds .
v.c . rubin and g.v .
coyne , ( princeton univ : princeton ) , p. 169 willick , j.a .
, courteau , s. , faber , s.m . , burstein , d. , & dekel , a. 1995 , apj , 446 , 12 willick , j.a . ,
courteau , s. , faber , s.m . ,
burstein , d. , dekel , a. , & kolatt , t. 1996 , apj , 457 , 460 yahil , a. , tammann , g.a . & sandage , a. 1977 , apj 217 , 903 | the recent re - evaulation of the peculiar velocity field outside the local supercluster ( dekel et al .
1999 ) has permitted a re - examination of the origins of the peculiar velocity field within the local supercluster using the mark iii catalog of galaxy peculiar velocities .
it is shown that the large - scale coherent pattern of peculiar velocities within the lsc are well - fit by a combination of the outside - region-(generated)-motions ( o - r - m ) from the potent model with a virgocentric infall pattern that produces 220 km s@xmath0 of virgocentric infall at the local group towards the virgo cluster moving at 88 km s@xmath0 towards the local group .
the part of the lg cmb motion this model can not fit is that perpendicular to the supergalactic plane . on what size scale
this sgz motion of the local group is shared by neighboring galaxies can not be determined from the present data set , but can be found if we can accurately measure galaxy distances close to the galactic plane . |
the process of growing films by the deposition of atoms on a substrate is of considerable experimental and theoretical interest @xcite .
while there has been a lot of research on the process of kinetic roughening @xcite leading to a self - affine interface profile , there has been much recent experimental @xcite and theoretical @xcite interest in a different mode of surface growth involving the formation of `` mounds '' which are pyramid - like or `` wedding - cake - like '' structures .
the precise experimental conditions that determine whether the growth morphology would be kinetically rough or dominated by mounds are presently unclear .
however , many experiments show the formation of mounds that _ coarsen _ ( the typical lateral size of the mounds increases ) with time . during this process , the typical slope of the sides of the pyramid - like mounds may or may not remain constant . if the slope remains constant in time , the system is said to exhibit _
slope selection_. as the mounds coarsen , the surface roughness characterized by the root - mean - square width of the interface increases .
eventually , at very long times , the system is expected to evolve to a single - mound structure in which the mound size is equal to the system size .
there are obvious differences between the structures of kinetically rough and mounded interfaces . in the first case ,
the interface is rough in a self - affine way at length scales shorter than a characteristic length @xmath0 that initially increases with time and eventually saturates at a value comparable to the sample size . in the second case ,
the characteristic length is the typical mound size @xmath1 whose time - dependence is qualitatively similar to that of @xmath0 .
the interface in this case looks well - ordered at length scales shorter than @xmath1 . nevertheless , there are certain similarities between the gross features of these two kinds of surface growth .
first consider the simpler situation in which the slope of the sides of the mounds remains constant in time .
simple geometry tells us that if the system evolves to a single - mound structure at long times , then the `` roughness exponent '' @xmath2 must be equal to unity .
also , the height - difference correlation function @xmath3 is expected to be proportional to @xmath4 for @xmath5 .
this is consistent with @xmath6 .
if the mound size @xmath1 increases with time as a power law , @xmath7 , during coarsening , then the interface width @xmath8 , which is essentially the height of a typical mound , should also increase with time as a power law with the same exponent @xmath9 .
thus , dynamic scaling with `` growth exponent '' @xmath10 equal to @xmath9 , and `` dynamical exponent '' @xmath11 equal @xmath12 is recovered .
if the mound slope @xmath13 increases with time as a power law , @xmath14 ( this is known in the literature as _ steepening _ ) , then one obtains behavior similar to anomalous dynamical scaling @xcite with @xmath15 , @xmath16 .
these similarities between the gross scaling properties of kinetic roughening with a large value of @xmath2 and mound formation with power - law coarsening make it difficult to experimentally distinguish between these two modes of surface growth .
this poses a problem in the interpretation of experimental results @xcite .
existing experiments on mound formation show a wide variety of behavior . without going into the details of individual experiments
, we note that some experiments show mound coarsening with a time - independent `` magic '' slope , whereas other experiments do not show any slope selection .
the detailed morphology of the mounds varies substantially from one experiment to another .
the reported values of the coarsening exponent @xmath9 show a large variation in the range 0.15 - 0.4 .
traditionally , the formation of mounds has been attributed to the presence of the so - called ehrlich - schwoebel ( es ) step - edge barrier @xcite that hinders the downward motion of atoms across the edge of a step .
this step - edge diffusion bias makes it more likely for an atom diffusing on a terrace to attach to an ascending step than to a descending one .
this leads to an effective `` uphill '' surface current @xcite that has a destabilizing effect , leading to the formation of mounded structures as the atoms on upper terraces are prevented by the es barrier from coming down .
this destabilizing effect is usually represented in continuum growth equations by a _
linear instability_. such growth equations usually have a `` conserved '' form in which the time - derivative of the height is assumed to be equal to the negative of the divergence of a nonequilibrium surface current @xmath17 .
the effects of an es barrier are modeled in these equations by a term in @xmath17 that is proportional to the gradient of the height ( for small values of the gradient ) with a _ positive _ proportionality constant .
such a term is manifestly unstable , leading to unlimited exponential growth of the @xmath18 fourier components of the height .
this instability has to be controlled by other nonlinear terms in the growth equation in order to obtain a proper description of the long - time behavior .
a number of different choices for the nonlinear terms have been reported in the literature @xcite .
if the `` es part '' of @xmath17 has one or more stable zeros as a function of the slope @xmath19 , then the slope of the mounds that form as a result of the es instability is expected to stabilize at the corresponding value(s ) of @xmath19 at long times .
the system would then exhibit slope selection .
if , on the other hand , this part of @xmath17 does not have a stable zero , then the mounds are expected to continue to steepen with time .
analytic and numerical studies of such continuum growth equations have produced a wide variety of results , such as power - law coarsening and slope selection with @xmath20 @xcite or @xmath21 @xcite in two dimensions , power - law coarsening accompanied by a steepening of the mounds @xcite , and a complex coarsening process @xcite in which the growth of the mound size becomes extremely slow after a characteristic size is reached .
there are several atomistic , cellular - automaton - type models @xcite that incorporate the effects of an es diffusion barrier .
formation and coarsening of mounds in the presence of an es barrier have also been studied @xcite in a 1d model with both discrete and continuum features .
we also note that a new mechanism for mounding instability has been discovered recently @xcite .
this instability , generated by fast diffusion along the edges of monatomic steps , leads to the formation of quasi - regular shaped mounds in two or higher dimensions .
the effects of this instability have been studied in simulations @xcite .
the wide variety of results @xcite obtained from simulations of different models , combined with similar variations in the experimental results , have made it very difficult to identify the microscopic mechanism of mound formation in surface growth . in this paper , we show that mound formation , slope selection and power - law coarsening in a class of one - dimensional ( 1d ) continuum growth equations and discrete atomistic models can occur from a mechanism that is radically different from the ones mentioned above .
our study is based on the conserved nonlinear growth equation proposed by villain @xcite and by lai and das sarma @xcite , and an atomistic version @xcite of this equation .
we have studied the behavior of the continuum equation by numerical integration , and the behavior of the atomistic model by stochastic simulation .
previous work @xcite on these systems showed that they exhibit a _ nonlinear _ instability , in which pillars ( grooves ) with height ( depth ) greater than a critical value continue to grow rapidly .
this instability can be controlled @xcite by the introduction of an infinite number of higher - order gradient nonlinearities .
when the parameter that describes the effectiveness of control is sufficiently large , the controlled models exhibit @xcite kinetic roughening , characterized by usual dynamical scaling with exponent values close to those expected from dynamical renormalization group calculations @xcite . as the value of the control parameter
is decreased , these models exhibit transient multiscaling @xcite of height fluctuations . for
yet smaller values of the control parameter , the rapid growth of pillars or grooves causes a breakdown of dynamical scaling , with the width versus time plot showing a sharp upward deviation @xcite from the power - law behavior found at short times ( before the onset of the nonlinear instability ) .
we report here the results of a detailed study of the behavior of these models in the regime of small values of the control parameter where conventional kinetic roughening is not observed .
we find that in this regime , the interface self - organizes into a sawtooth - like structure with a series of triangular , pyramid - like mounds .
these mounds coarsen in time , with larger mounds growing at the expense of smaller ones . in this coarsening regime ,
a power - law dependence of the interface width on time is recovered .
the slope of the mounds remains constant during the coarsening process . in section [ models ] , the growth equation and the atomistic model studied in this work are defined and the numerical methods we have used to analyze their behavior are described .
the basic phenomenology of mound formation and slope selection in these systems is described in detail in section [ mound ] .
specifically , we show that the nonlinear mechanism of mound formation in these systems is `` generic '' in the sense that the qualitative behavior does not depend on the specific form of the function used for controlling the instability .
in particular , we find very similar behavior for two different forms of the control function : one used in earlier studies @xcite of these systems , and the other one proposed by politi and villain @xcite from physical considerations . since the linear instability used conventionally to model the es mechanism is explicitly absent in our models , our work shows that the presence of step - edge barriers is not essential for mound formation .
the slope selection found in our models is a true example of nonlinear pattern formation : since the nonequilibrium surface current in our models vanishes for all values of constant slope , the selected value of the slope can not be predicted in any simple way .
this is in contrast to the behavior of es - type models where slope selection occurs only if the surface current vanishes at a specific value of the slope .
next , in section [ pt ] , we show that the change in the dynamical behavior of the system ( from kinetic roughening to mound formation and coarsening ) may be described as a first - order nonequilibrium phase transition .
since the instability in our models is a nonlinear one , the flat interface is locally stable in the absence of noise for all values of the model parameters ( the strength of the nonlinearity and the value of the control parameter ) .
the mounded phase corresponds to a different stationary solution of the dynamical equations in the absence of noise .
we use a linear stability analysis to find the `` spinodal boundary '' in the two - dimensional parameter space across which the mounded stationary solution becomes locally unstable .
we show that the results of this numerical stability analysis can also be obtained from simple analytic arguments . to obtain the phase boundary in the presence of noise ,
we first define an _ order parameter _ that is zero in the kinetically rough phase and nonzero in the mounded phase .
we combine the numerically obtained results for this order parameter for different sample sizes with finite - size scaling to confirm that this order parameter exhibits the expected behavior in the two phases .
the phase boundary that separates the mounded phase from the kinetically rough one is obtained numerically .
the phase boundaries for the continuum model with two different forms of the control function and the atomistic model are found to be qualitatively similar .
the results of a detailed study of the process of coarsening of the mounds are reported in section [ coars ] .
surprisingly , we find that the coarsening exponents of the continuum equation and its atomistic version are different .
we propose a possible explanation of this result on the basis of an analysis of the coarsening process in which the problem is mapped to that of of a brownian walker in an attractive force field . in this mapping ,
the brownian walk is supposed to describe the noise - induced random motion of the peak of a mound , and the attractive `` force '' represents the interaction between neighboring mounds that leads to coarsening .
we show that the numerical results obtained for the dynamics of mounds in the atomistic model are consistent with this explanation . in section [ conserv ]
, we consider the behavior of the continuum growth equation for `` conserved '' noise statistics .
the nonlinear instability found in the nonconserved case is expected to be present in the conserved case also .
however , there is an important difference between the two cases .
the nonconserved model exhibits anomalous dynamical scaling , so that the typical nearest - neighbor height difference continues to increase with time , and the instability is always reached @xcite at sufficiently long times , even if the starting configuration is perfectly flat . since the continuum model with conserved noise statistics exhibits @xcite usual dynamic scaling with @xmath22 , the nearest - neighbor height difference is expected to saturate at long times if the initial state of the system is flat . under these circumstances ,
the occurrence of the nonlinear instability in runs started from flat states would depend on the values of the parameters .
specifically , the instability may not occur at all if the value of the nonlinear coefficient in the growth equation is sufficiently small .
at the same time , the instability can be initiated by choosing an initial state with sufficiently high ( deep ) pillars ( grooves ) .
since mound formation in these models is crucially dependent on the occurrence of the instability , the arguments above suggest that the nature of the long - time steady state reached in the conserved model may depend on the choice of the initial state .
indeed , we find from simulations that in a region of parameter space , the mounded and kinetically rough phases are both locally stable and the steady state configuration is determined by the choice of the initial configuration of the interface .
these results imply the surprising conclusion that the long - time , steady - state morphology of a growing interface , as well as the dynamics of the process by which the steady state is reached may be `` history dependent '' in the sense that the behavior would depend crucially on the choice of the initial state .
a summary of our findings and a discussion of the implications of our results are provided in sec.[summ ] .
a summary of the basic results of our study was reported in a recent letter @xcite .
conserved growth equations ( deterministic part of the dynamics having zero time derivative for the @xmath23 fourier mode of the height variable ) with nonconserved noise are generally used @xcite to model nonequilibrium surface growth in molecular beam epitaxy ( mbe ) .
the conservation is a consequence of absence of bulk vacancies , overhangs and desorption ( evaporation of atoms from the substrate ) under optimum mbe growth conditions .
thus , integrating over the whole sample area gives the number of particles deposited .
this conservation is not strictly valid because of `` shot noise '' fluctuations in the beam .
the shot noise is modeled by an additive noise term @xmath24 in the equation of motion of the interface .
the noise @xmath25 is generally assumed to be delta - correlated in both space and time : @xmath26 where @xmath27 is a point on a @xmath28-dimensional substrate .
thus , a conserved growth equation may be written in a form @xmath29 where @xmath30 is the height at point @xmath31 at time @xmath32 , and @xmath33 is the surface current density . the surface current models the deterministic dynamics at the growth front .
as mentioned in section [ intro ] , the presence of an es step - edge barrier is modeled in continuum equations of the form of eq.([mbe ] ) by a term in @xmath17 that is proportional to the slope @xmath34 , with a positive constant of proportionality .
this makes the flat surface ( @xmath35 constant for all @xmath27 ) linearly unstable .
this instability is controlled by the introduction of terms involving higher powers of the local slope @xmath19 and higher - order spatial derivatives of @xmath36 .
we consider the conserved growth equation proposed by villain @xcite and lai and das sarma @xcite for describing mbe - type surface growth _ in the absence of es step - edge barriers_. this equation is of the form @xmath37 where @xmath38 represents the height variable at the point @xmath27 at time @xmath39 .
this equation is believed @xcite to provide a correct description of the kinetic roughening behavior observed in mbe - type experiments @xcite . in our study
, we numerically integrate the 1d version of eq.([lds1 ] ) using a simple euler scheme @xcite . upon choosing appropriate units of length and time and discretizing in space and time , eq.([lds1 ] ) is written as @xcite @xmath40 \nonumber \\ & + & \sqrt{\delta t}\ , \eta_i(t ) , \label{lds2}\end{aligned}\ ] ] where @xmath41 represents the dimensionless height variable at the lattice point @xmath42 at dimensionless time @xmath32 , @xmath43 and @xmath44 are lattice versions of the derivative and laplacian operators , and @xmath45 is a random variable with zero average and variance equal to unity .
these equations , with an appropriate choice of @xmath46 , are used to numerically follow the time evolution of the interface . in most of our studies
, we have used the following definitions for the lattice derivatives : @xmath47 we have checked that the use of more accurate , left - right symmetric definitions of the lattice derivatives , involving more neighbors to the left and to the right @xcite , leads to results that are very similar to those obtained from calculations in which these simple definitions are used .
we have also checked that the results obtained in the deterministic limit ( @xmath48 ) by using a more sophisticated integration routine @xcite closely match those obtained from the euler method with sufficiently small values of the integration time step .
we have also studied an atomistic version @xcite of eq.([lds1 ] ) in which the height variables @xmath49 are integers .
this model is defined by the following deposition rule .
first , a site ( say @xmath42 ) is chosen at random .
then the quantity @xmath50 is calculated for the site @xmath42 and all its nearest neighbors .
then , a particle is added to the site that has the smallest value of @xmath51 among the site @xmath42 and its nearest neighbors . in the case of a tie for the smallest value
, the site @xmath42 is chosen if it is involved in the tie ; otherwise , one of the sites involved in the tie is chosen randomly .
the number of deposited layers provides a measure of time in this model .
it was found in earlier studies @xcite that both these models exhibit a _ nonlinear instability _ in which isolated structures ( pillars for @xmath52 , grooves for @xmath53 ) grow rapidly if their height ( depth ) exceeds a critical value .
this instability can be controlled @xcite by replacing @xmath54 in eqns.([lds2 ] ) and ( [ kds1 ] ) by @xmath55 where the nonlinear function @xmath56 is defined as @xmath57 @xmath58 being a control parameter .
we call the resulting models `` model i '' and `` model ii '' , respectively .
this replacement , amounts to the introduction of an infinite series of higher - order nonlinear terms .
the time evolution of the height variables in model i is , thus , given by @xmath59 + \sqrt{\delta t}\ , \eta_i(t ) .
\label{cld}\end{aligned}\ ] ] in model ii , the quantity @xmath60 is defined as @xmath61 while the function @xmath56 was introduced in the earlier work purely for the purpose of controlling the nonlinear instability , it turns out that the introduction of this function in the growth equation is physically meaningful .
politi and villain @xcite have shown that the nonequilibrium surface current that leads to the @xmath62 term in eq.([lds1 ] ) should be proportional to @xmath63 when @xmath64 is small , and should go to zero when @xmath64 is large .
the introduction of the `` control function '' @xmath55 satisfies this physical requirement .
we have also carried out studies of a slightly different model ( which we call `` model ia '' ) in which the function @xmath56 is assumed to have a form suggested by politi and villain : @xmath65 where @xmath66 is , as before , a positive control parameter .
this function has the same asymptotic behavior as that of the function defined in eq.([fofx ] ) .
as we shall show later , the results obtained from calculations in which these two different forms of @xmath56 are used are qualitatively very similar .
in fact , we expect that the qualitative behavior of these models would be the same for any monotonic function @xmath56 that satisfies the following requirements : ( i ) @xmath56 must be proportional to @xmath67 in the small-@xmath67 limit ; and ( ii ) it must saturate to a constant value as @xmath68 . we have carried out extensive simulations of both these models for different system sizes .
the results reported here have been obtained for systems of sizes @xmath69 .
there is no significant dependence of the results on @xmath70 .
the time step used in most of our studies of models i and ia is @xmath71 .
we have checked that very similar results are obtained for smaller values of @xmath46 .
we used both unbounded ( gaussian ) and bounded distributions for the random variables @xmath72 in our simulations of models i and ia , with no significant difference in the results . for computational convenience , a bounded distribution ( uniform between @xmath73 and @xmath74 )
was used in most of our calculations .
unless otherwise stated , the results described in the following sections were obtained using periodic boundary conditions .
the effects of using other boundary conditions will be discussed in the next section .
it has been demonstrated earlier @xcite that if the control parameter @xmath66 is sufficiently large , then the nonlinear instability is completely suppressed and the models exhibit the usual dynamical scaling behavior with the expected @xcite exponent values , @xmath75 , @xmath76 , and @xmath77 .
this behavior for model i is illustrated by the solid line in fig.[bigfig1 ] , which shows a plot of the width @xmath8 as a function of time @xmath32 for parameter values @xmath78=@xmath79 and @xmath80 . as the value of @xmath66
is decreased with @xmath78 held constant , the instability makes its appearance : the height @xmath81 of an isolated pillar ( for @xmath52 ) increases in time if @xmath82 . the value of @xmath83 is nearly independent of @xmath66 , while @xmath84 increases as @xmath66 is decreased @xcite . if @xmath66 is sufficiently large , @xmath84 is small and the instability does not affect the scaling behavior of global quantities such as @xmath8 , although transient multiscaling at length scales shorter than the correlation length @xmath85 may be found @xcite if @xmath66 is not very large . as @xmath66 is decreased further , @xmath84 becomes large , and when isolated pillars with @xmath86 are created at an initially flat interface through fluctuations , the rapid growth of such pillars to height @xmath84 leads to a sharp upward departure from the power - law scaling of @xmath8 with time @xmath32 .
the time at which this departure occurs varies from run to run .
this behavior for model i with @xmath87 and @xmath88 is shown by the dash - dotted line in fig.[bigfig1 ] .
this instability leads to the formation of a large number of randomly distributed pillars of height close to @xmath84 .
as the system evolves in time , the interface self - organizes to form triangular mounds of a fixed slope near these pillars .
these mounds then coarsen in time , with large mounds growing larger at the expense of small ones . in this coarsening regime , a power - law growth of @xmath8 in time is recovered .
the slope of the sides of the triangular mounds remains constant during this process . finally , the system reaches a steady state with one peak and one trough ( if periodic boundary conditions are used ) and remains in this state for longer times .
the interface profiles in the kinetically rough phase ( obtained for relatively large values of @xmath66 ) and the mounded phase ( obtained for small @xmath66 ) are qualitatively different .
this difference is illustrated in fig.[bigfig2 ] that shows typical interface profiles in the two different phases .
this figure also shows a typical interface profile for model ia in the mounded regime , illustrating the fact that the precise choice of the control function @xmath56 is not crucial for the formation of mounds .
the evolution of the interface structure in the mounded regime of model i is illustrated in fig.[bigfig3 ] which shows the interface profiles obtained in a typical @xmath89 run starting from a flat initial state at three different times : @xmath90 ( before the onset of the instability ) , @xmath91 ( after the onset of the instability , in the coarsening regime ) , and @xmath92 ( in the final steady state ) .
this figure also shows the steady - state interface profile of a @xmath93 sample with the same parameters , to illustrate that the results do not depend on the sample size .
very similar behavior is found for model ii . since the heights in this atomistic model can increase only by discrete amounts in each unit of discrete time ,
the increase of @xmath8 at the onset of the instability is less rapid here than in the continuum models i and ia .
nevertheless , the occurrence of the instability for small values of @xmath66 shows up as a sharp upward deviation of the @xmath8 versus @xmath32 plot from the initial power - law behavior with @xmath75 .
this is illustrated by the dash - dotted line in fig.[bigfig4 ] , obtained from simulations of model ii with @xmath94 , @xmath95 .
this behavior is to be contrasted with that for @xmath94 , @xmath96 , shown by the full line in fig.[bigfig4 ] , where the nonlinear instability is absent .
the difference between the surface morphologies in the two regimes of model ii is illustrated in fig.[bigfig5 ] . the kinetically rough , self - affine morphology obtained for @xmath88 is clearly different from the mounded profile found for @xmath95 .
the time evolution of the interface in the mounded regime of this model is illustrated in fig.[bigfig6 ] .
the general behavior is clearly similar to that found for models i and ia .
this figure also shows a properly scaled plot of the interface profile of a @xmath93 sample with the same parameters at a time in the coarsening regime .
it is clear from this plot that the nature of the interface and the value of the selected slope do not depend on the sample size .
the occurrence of a peak and a symmetrically placed trough in the steady - state profiles shown in figs [ bigfig3 ] and [ bigfig6 ] is a consequence of using periodic boundary conditions .
the deterministic part of the growth equation of eq.([cld ] ) strictly conserves the average height if periodic boundary conditions are used .
so , the average height remains very close to zero if the initial state is flat , as in most of our simulations .
the steady - state profile must have at least one peak and one trough in order to satisfy this requirement . also , it is easy to show that if the slopes of the `` uphill '' and `` downhill '' parts of the steady - state profile are the same in magnitude ( this is true for our models ) , then the two extrema must be separated by @xmath97 .
therefore , it is clear that the steady state obtained in simulations with periodic boundary conditions must have a peak and a symmetrically placed trough separated by distance @xmath97 . to check whether the basic phenomenology described above depends on the choice of the boundary condition ,
we have carried out test simulations using two other boundary conditions : `` fixed '' boundary condition , in which the height variable to the left of @xmath98 , and to the right of @xmath99 are pinned to zero at all times ; and `` zero flux '' boundary condition with vanishing first and third derivatives of the height at the two ends of the sample .
for these boundary conditions , the deterministic part of the growth equation does not strictly conserve the average height . as a result , the symmetry between the mound and the trough , found in the long - time steady state obtained for periodic boundary condition , is not present if one of the other boundary conditions is used . in particular , it is possible to stabilize a single mound or a single trough in the steady state for the other boundary conditions . since the heights at the two ends must be the same for fixed boundary condition , the two extrema in a configuration with one mound and one trough must be separated by @xmath97 , as shown in fig.[bigfig7 ] .
the two extrema would not be separated by @xmath97 for the zero - flux boundary condition .
these effects of boundary conditions are illustrated in fig.[bigfig7 ] which shows profiles in the mounded regime obtained for the three different boundary conditions mentioned above .
it is clear from the results shown in this figure that the basic phenomenology , i.e. the formation and coarsening of mounds and slope selection , is not affected by the choice of boundary conditions . in particular
, the values of the selected slope and the heights of the pillars at the top of a mound and the bottom of a trough remain unchanged when boundary conditions other than periodic are used .
the selection of a `` magic slope '' during the coarsening process is clearly seen in the plots of fig.[bigfig3 ] and fig.[bigfig6 ] .
more quantitatively , the probability distribution of the magnitude of the nearest - neighbor height differences @xmath100 is found to exhibit a pronounced peak at the selected value of the slope , and the position of this peak does not change during the coarsening process .
fig.[bigfig8 ] shows a comparison of the distribution of the magnitude of the nearest - neighbor height difference for model i in the mounded and kinetically rough phases .
a bimodal distribution is seen for the mounded phase , the two peaks corresponding to the values of the selected slope and the height of the pillars at the top and bottom of the pyramids .
the kinetically rough phase , on the other hand , exhibits a distribution peaked at zero .
fig.[bigfig9 ] shows the values of the selected slope at different times in the coarsening regime of model i. the constancy of the slope is clearly seen in this plot .
all these features remain true for the discrete model .
plots of the distribution @xmath101 at two different times in the coarsening regime of model ii are shown in fig.[bigfig10 ] .
the peak position shows a small shift in the positive direction as @xmath32 is increased , but this shift is small compared to the width of the distributions , indicating near constancy of the selected slope .
the value of the selected slope depends on the parameters @xmath78 and @xmath66 .
this is discussed in the next section .
the instability that leads to mound formation in our models is a nonlinear one , so that the perfectly flat state of the interface is a locally stable steady - state solution of the zero - noise growth equation for all parameter values .
when the instability is absent ( e.g. for large values of the control parameter @xmath66 ) , this `` fixed - point '' solution of the noise - free equation is transformed to the kinetically rough steady state in the presence of noise .
the mounded steady state obtained for small values of @xmath66 must correspond to a different fixed point of the zero - noise growth equation .
such nontrivial fixed - point solutions may be obtained from the following simple calculation .
the profile near the top ( @xmath102 ) of a triangular mound may be approximated as @xmath103 , where @xmath104 is the height of the pillar at the top of the mound and @xmath105 is the selected slope .
this profile would not change under the dynamics of eq.([cld ] ) with no noise if the following conditions are satisfied : @xmath106 these conditions lead to the following pair of non - linear equations for the variables @xmath104 and @xmath105 used to parametrize the profile near the top of a mound : @xmath107/c & = & 0 , \nonumber \\ 3x_1 - x_2 - \lambda [ 1-e^{-c(x_1+x_2)^2/4}]/c & = & 0 . \label{stability}\end{aligned}\ ] ] these equations admit a non - trivial solution for sufficiently small @xmath66 , and the resulting values of @xmath104 and @xmath105 are found to be quite close to the results obtained from numerical integration . a similar analysis for the profile near the bottom of a trough ( this amounts to replacing @xmath105 by @xmath108 in eq.([stability ] ) ) yields slightly different values for @xmath104 and @xmath105 . the full stable profile ( a fixed point of the dynamics without noise ) with one peak and one trough
may be obtained numerically by calculating the values of @xmath109 for which @xmath110 , the term multiplying @xmath46 in the right - hand side of eq.([cld ] ) , is zero for all @xmath42 .
the fixed - point values of @xmath109 satisfy the following equations : @xmath111 = 0\,\ , \hbox{for\,\,all}\,\,i.\label{fp}\ ] ] a numerical solution of these coupled nonlinear equations shows that the small mismatch between the values of @xmath105 near the top and the bottom is accommodated by creating a few ripples near the top .
the numerically obtained fixed - point profile for a @xmath93 system with @xmath87 , @xmath88 is shown in fig.[bigfig11 ] , along with a typical steady - state profile for the same system .
the two profiles are found to be nearly identical , indicating that the mounded steady state in the presence of noise corresponds to this fixed - point solution of the noiseless discretized growth equation .
fixed - point solutions of the continuum equation , eq.([lds1 ] ) , with @xmath112 and @xmath113 replaced by @xmath114 where @xmath56 has the form shown in eq.([pvform ] ) may also be obtained by a semi - analytical approach following racz _
et al . _
we consider stationary solutions of the continuum equation that satisfy the following first - order differential equation with appropriate boundary conditions : @xmath115 where @xmath116 is the local slope of the interface and @xmath117 is a constant that must be positive in order to obtain a solution that resembles a triangular mound . at large distances from the peak of the mound ,
the slope @xmath118 would be constant , so that @xmath119 would vanish , whereas the second term would give a positive contribution if @xmath78 is positive . at the peak of the profile ,
the second term would be zero because @xmath118 is zero , but @xmath119 would be negative , making the left - hand side of eq.([cont ] ) positive .
while a closed - form solution of this differential equation can not be obtained , the value of @xmath120 at any point @xmath67 may be calculated with any desired degree of accuracy by numerically solving a simple algebraic equation .
the height profile is then obtained by integrating @xmath120 with appropriate boundary conditions . in our calculation
, we used the procedure of racz _ et .
@xcite to take into account periodic boundary conditions . in fig .
[ bigfig11 ] , we have shown a typical steady - state profile of a @xmath89 sample of model ia with @xmath87 and @xmath121 , and a fixed - point solution of the corresponding continuum equation . the value of the constant @xmath117 in eq.([cont ] )
was chosen to yield the same slope as that of the steady - state profile of the discrete model .
these results show that the steady - state properties for the two forms of @xmath56 are very similar , and the continuum equation admits stationary solutions that are very similar to those of the discretized models .
the local stability of the mounded fixed point may be determined from a calculation of the eigenvalues of the stability matrix , @xmath122 , evaluated at the fixed point .
we find that the largest eigenvalue of this matrix ( disregarding the trivial zero eigenvalue associated with an uniform displacement of the interface , @xmath123 for all @xmath42 ) crosses zero at @xmath124 ( see fig.([bigfig12 ] ) ) , signaling an instability of the mounded profile .
the structure of eq.([cld ] ) implies that @xmath125 .
thus , for @xmath126 , the dynamics of eq.([cld ] ) without noise admits two locally stable invariant profiles : a trivial , flat profile with @xmath127 the same for all @xmath42 , and a non - trivial one with one mound and one trough .
depending on the initial state , the noiseless dynamics takes the system to one of these two fixed points .
for example , an initial state with one pillar on a flat background is driven by the noiseless dynamics to the flat fixed point if the height of the pillar is smaller than a critical value , and to the mounded one otherwise .
the `` relevant '' perturbation that makes the mounded fixed point unstable at @xmath128 is a uniform vertical relative displacement of the segment of the interface between the peak and the trough of the fixed - point profile .
this can be seen by numerically evaluating the right eigenvector corresponding to the eigenvalue of the stability matrix that crosses zero at @xmath128 .
this is demonstrated in the inset of fig.[bigfig12 ] . also ,
examination of the time evolution of the mounded structure for values of @xmath66 slightly higher than @xmath129 shows that the instability of the structure first appears at the bottom of the trough . taking cue from these observations
, the value @xmath129 can be obtained from a simple calculation .
we consider the profile near the bottom of a trough at @xmath102 .
as discussed above , the profile near @xmath130 may be parametrized as @xmath131 , @xmath132 , and the values of @xmath104 and @xmath105 may be obtained by solving a pair of nonlinear equations , eq.([stability ] ) with @xmath105 replaced by @xmath108 .
we now consider a perturbation of this profile , in which the heights on one side of @xmath130 are all increased by a small amount @xmath133 ( i.e. @xmath134 , @xmath135 with @xmath136 ) , and use eq.([cld ] ) to calculate how @xmath133 changes with time , assuming its value to be small .
the requirement that @xmath133 must decrease with time for the fixed - point structure to be locally stable leads to the following equation for the value of @xmath66 at which the structure becomes unstable : @xmath137 by substituting the numerically obtained values of @xmath104 and @xmath105 in this equation , the critical value , @xmath138 , of the parameter @xmath66 is obtained as a function of @xmath78 .
the values obtained this way are in good agreement with those obtained from our full numerical calculation of the eigenvalues of the stability matrix .
the `` spinodal '' lines ( i.e. the lines in the @xmath139 plane beyond which the mounded fixed point is unstable ) for models i and ia are shown in fig.[bigfig13 ] .
both these lines have the expected form , @xmath140 .
it would be interesting to carry out a similar stability analysis for the mounded stationary profile ( see fig.[bigfig11 ] ) of the continuum equation corresponding to model ia .
such a calculation would have to be performed _ without discretizing space _ if we want to address the question of whether the behavior of the truly continuum equation is similar to that of the discretized versions considered here .
we have not succeeded in carrying out such a calculation : since the mounded stationary profiles for the continuum equation are obtained from a numerical calculation , it would be extremely difficult , if not impossible , to carry out a linear stability analysis for such stationary solutions without discretizing space . in the presence of the noise
, the perfectly flat fixed point transforms to the kinetically rough steady state , and the non - trivial fixed point evolves to the mounded steady state shown in fig.[bigfig11 ] .
a dynamical phase transition at @xmath141 separates these two kinds of steady states . to calculate @xmath142
, we start a system at the mounded fixed point and follow its evolution according to eq.([cld ] ) for a long time ( typically @xmath143 ) to check whether it reaches a kinetically rough steady state . by repeating this procedure many times , the probability , @xmath144 , of a transition to a kinetically rough state
is obtained . for fixed @xmath78 , @xmath145 increases rapidly from 0 to 1 as @xmath66 is increased above a critical value .
typical results for @xmath145 as a function of @xmath66 for model i with @xmath87 are shown in the inset of fig.[bigfig13 ] . the value of @xmath66 at which @xmath146 provides an estimate of @xmath147 .
another estimate is obtained from a similar calculation of @xmath148 , the probability that a flat initial state evolves to a mounded steady state . as expected
, @xmath149 increases sharply from 0 to 1 as @xmath66 is decreased ( see inset of fig.[bigfig13 ] ) , and the value of @xmath66 at which this probability is 0.5 is slightly lower than the value at which @xmath146 .
this difference reflects finite - time hysteresis effects .
the value of @xmath147 is taken to be the average of these two estimates , and the difference between the two estimates provides a measure of the uncertainty in the determination of @xmath147 .
the phase boundary obtained this way is shown in fig.[bigfig13 ] , along with the results for @xmath142 obtained for the discrete model ii from a similar analysis .
the general behavior found for all the models as the parameters @xmath78 and @xmath66 are varied is qualitatively very similar to that in equilibrium first order phase transitions of two- and three - dimensional systems as the temperature and other parameters , such as the magnetic field in spin systems , are varied . to take a standard example of an equilibrium first order transition
, we consider a system with a scalar order - parameter field @xmath150 , described by a ginzburg - landau free energy functional @xcite that has a cubic term : @xmath151 = \int d{\bf r } \left[\frac{1}{2 } a \psi^2({\bf r } ) - \frac{1}{3 } b \psi^3({\bf r } ) + \frac{1}{4 } u \psi^4({\bf r})\right ] , \label{gl}\ ] ] where @xmath152 and @xmath153 are positive constants , @xmath154 with @xmath155 , and @xmath156 is the temperature .
considering uniform states , @xmath157 , the free energy per unit volume may be written as @xmath158 it is easy to show that for @xmath159 , the function @xmath160 has two local minima , one at @xmath161 , and the other at a positive value of @xmath162 .
these two minima represent the two phases of the system .
this system exhibits a first order equilibrium phase transition from the disordered phase ( @xmath161 ) to an ordered phase with positive @xmath162 as the temperature is decreased .
the transition temperature @xmath163 lies between @xmath164 and @xmath165 .
the temperature @xmath165 at which the minimum corresponding to the ordered phase disappears is called the `` spinodal '' temperature for the ordered phase .
the spinodal temperature for the disordered phase is @xmath164 .
now consider the dynamics of this system according to the following time - dependent ginzburg - landau equation @xcite : @xmath166 } { \delta \psi({\bf r},t ) } + \eta({\bf r},t ) , \label{tdgl}\ ] ] where @xmath167 is a kinetic coefficient and @xmath25 represents gaussian delta - correlated noise whose variance is related to @xmath167 and the temperature @xmath156 via the fluctuation - dissipation theorem @xcite . in the absence of noise
, this equation converges to local minima of the functional @xmath168 .
so , the noiseless dynamics exhibits two locally stable fixed points for @xmath169 , corresponding to the two minima of @xmath160 that represent the disordered and uniformly ordered states .
this is analogous to the two locally stable fixed points of our nonequilibrium dynamical systems for @xmath170 .
if we identify the flat and mounded fixed points as the `` disordered '' and `` ordered '' states , respectively , and the control parameter @xmath66 to play the role of the temperature @xmath156 , then the noiseless dynamics of our models would look similar to that of eq.([tdgl ] ) for @xmath171 , with @xmath129 playing the role of the spinodal temperature @xmath165 of the equilibrium problem . in the presence of noise , the system described by eq.([tdgl ] )
exhibits a first - order phase transition at @xmath163 that lies between @xmath165 and @xmath164 : the system selects one of the phases corresponding to the two fixed points of the noiseless dynamics , except at @xmath163 where both phase coexist .
the local stability of the mean - field ordered and disordered states in a small temperature - interval around @xmath163 is manifested in the dynamics as finite - time hysteresis effects .
the behavior we find for out nonequilibrium dynamical models is qualitatively similar : the system selects the steady state corresponding to the mounded ( `` ordered '' ) fixed point of the noiseless dynamics as the control parameter @xmath66 ( analogous to the temperature @xmath156 of the equilibrium system ) is decreased below @xmath147 which is smaller than the `` spinodal '' value @xmath129 .
the growth models do not exhibit a `` spinodal '' point for the kinetically rough ( `` disordered '' ) phase : the flat fixed point of the noiseless dynamics is locally stable for all positive values of the control parameter @xmath66 .
if this analogy with equilibrium first order transition is correct , then our models should show hysteresis and coexistence of kinetically rough and mounded morphologies for values of @xmath66 near @xmath142 . as mentioned above
, we do find hysteresis ( see inset of fig.[bigfig13 ] ) in finite - time simulations with values of @xmath66 near @xmath147 .
evidence for two - phase coexistence is presented in fig.[bigfig14 ] , where a snapshot of the interface profile for a @xmath93 sample of model i with @xmath87 , @xmath172 is shown .
this value of @xmath66 is very close to the critical value @xmath147 for @xmath87 ( see inset of fig.[bigfig13 ] ) .
this plot clearly illustrates the simultaneous presence of mounded and rough morphologies in the interface profile .
the results described above suggest that our growth models exhibit a _ first - order dynamical phase transition _ at @xmath173 . to make this conclusion more concrete , we need to define an _ order parameter _ , analogous to the quantity @xmath162 in the equilibrium problem discussed above , that is zero in the kinetically rough phase , and jumps to a non - zero value as the system undergoes a transition to the mounded phase at @xmath174 .
the identification of such an order parameter would also be useful for distinguishing between these two different kinds of growth in experiments as mentioned in the introduction , it is difficult to experimentally differentiate between kinetic roughening and mound formation with coarsening from measurements of the usual bulk properties of the interface . a clear distinction between the two morphologies may be obtained from measurements of the average number of extrema of the height profile @xcite .
the steady - state profile in the mound - formation regime exhibits two extrema for _ all _ values of the system size @xmath70 .
in contrast , the number of extrema in the steady state in the kinetic roughening regime increases with @xmath70 as a power law @xcite we find that for values of @xmath66 for which the system is kinetically rough , e.g. for @xmath175 , @xmath176 for model i , the average number of extrema in the steady state is proportional to @xmath177 with @xmath178 .
this observation allows us to define an `` order parameter '' that is zero in the large-@xmath66 , kinetic roughening regime and finite in the small-@xmath66 , mound - formation regime .
let @xmath179 be an ising - like variable , equal to the sign of the slope of the interface at site @xmath42 .
an extremum in the height profile then corresponds to a `` domain wall '' in the configuration of the @xmath180 variables .
since there are two domain walls separated by @xmath181 in the steady state in the mound - formation regime , the quantity @xmath182 where @xmath183 represents a time - average in the steady state , would be finite in the @xmath184 limit .
on the other hand , @xmath162 would go to zero for large @xmath70 in the kinetically rough regime because the number of domains in the steady - state profile would increase with @xmath70 .
we find numerically that in the kinetically rough phase , @xmath185 with @xmath186 .
the finite - size scaling data for the order parameter @xmath162 for models i and ii for both faceted and kinetically rough phases is shown in fig.[bigfig15 ] .
it is seen that @xmath187 varies linearly with the system size @xmath70 in the mounded phase , whereas @xmath188 with @xmath186 for model i and @xmath189 for model ii in the the kinetically rough phase .
so , in the @xmath184 limit , the order parameter would jump from zero to a value close to unity as @xmath66 is decreased below @xmath142 .
this is exactly the behavior expected at a first - order phase transition .
the occurrence of a first - order phase transition in our 1d models with short - range interactions may appear surprising
it is well - known @xcite that 1d systems with short - range interactions do not exhibit any equilibrium thermodynamic transition at a non - zero temperature .
the situation is , however , different for nonequilibrium phase transitions : in contrast to equilibrium systems , a first - order phase transition may occur in one - dimensional nonequilibrium systems with short - range interactions .
several such transitions have been well documented in the literature @xcite .
so , there is no reason to _ a priori _ rule out the occurrence of a true first - order transition in our 1d nonequilibrium systems . as discussed above , our numerical results strongly suggest the existence of a true phase transition .
however , since all our results are based on finite - time simulations of finite - size systems , we can not claim to have established rigorously the occurrence of a true phase transition in our models .
the crucial question in this context is whether the order parameter @xmath162 would be nonzero in the mounded phase in the @xmath184 limit if the time - average in eq.([op ] ) is performed over arbitrarily long times .
since the steady - state profile in this phase has a single mound and a single trough ( this is clear from our simulations ) , the only way in which @xmath162 can go to zero is through strong `` phase fluctuations '' corresponding to lateral shifts of the positions of the peak and the trough .
we do not find any evidence for such strong phase fluctuations .
we have calculated the time autocorrelation function of the phase of the order parameter for small samples over times of the order of @xmath190 and found that it remains nearly constant at a value close to unity over the entire range of time .
so , if such phase fluctuation eventually make the order parameter zero for all values of @xmath66 , then this must happen over astronomically long times .
our finite - time simulations can not , of course , rule out this possibility .
during the late - stage evolution of the interface , the mounds coarsen with time , increasing the typical size of the triangular pyramidal structures .
the process of coarsening occurs by larger mounds growing larger at the expense of the smaller ones while always retaining their `` magic '' slope .
snapshots of the system in the coarsening regime are shown in figs [ bigfig16 ] and [ bigfig17 ] for model i and model ii , respectively .
the constancy of the slope during the coarsening process is clearly seen in these figures . as discussed in the introduction
, the constancy of the slope implies that if the typical lateral size of a mound increases in time as a power law with exponent @xmath9 ( @xmath191 ) , then the width of the interface would also increase in time as a power law with the same exponent ( @xmath192 with @xmath193 ) .
therefore , the value of the coarsening exponent @xmath9 may be obtained by measuring the width @xmath8 as a function of time in the coarsening regime . in fig.[bigfig18 ] , we show a plot of the width as a function of time for model i with @xmath194 , @xmath88 .
it is clear from the plot that the time - dependence of the width is well - described by a power law with @xmath195 .
a similar plot for the discrete model ii with @xmath94 , @xmath95 , shown in fig.[bigfig19 ] , also shows a power - law growth of the width in the long - time regime , but the value of the coarsening exponent obtained from a power - law fit to the data is @xmath196 , which is clearly different for the value obtained for model i. this is a surprising result : model ii was originally defined @xcite with the specific purpose of obtaining an atomistic realization of the continuum growth equation of eq.([lds1 ] ) , and earlier studies @xcite have shown that the dynamical scaling behavior of this model in the kinetic roughening regime is the same as that of model i. also , we have found in the present study that the dynamical phase transition in this model has the same character as that in model i. so , the difference in the values of the coarsening exponents for these two models is unexpected . as noted earlier , there is some evidence suggesting that the typical slope of the mounds in model ii increases very slowly with time ( see fig.[bigfig10 ] ) .
however , this `` steepening '' , if it actually occurs , is too slow to account for the large difference between the values of the coarsening exponents for models i and ii . in order to understand these numerical results , we first address the question of why the mounds coarsen with time .
this problem has certain similarities with domain growth in spin systems @xcite . using the ising variables @xmath180 defined in the preceding section
, each height profile can be mapped to a configuration of ising spins .
the coarsening of mounds then corresponds to a growth of the typical size of domains of these ising spins .
there is , however , an important difference between the coarsening of mounds in our models and the usual domain growth problem @xcite for ising spins .
domain growth in spin systems is the process through which the system approaches equilibrium from an out - of - equilibrium initial state .
the dynamics of this process may be understood in terms of arguments based on considerations of the free energy ( at finite temperatures ) or energy ( at zero temperature ) .
such arguments do not apply to our nonequilibrium growth models .
the reason for the coarsening of mounds in our models must be sought in the relative stability of different structures under the assumed dynamics and the effects of noise . as discussed in the preceding section , the fixed point of eq.([cld ] ) with one mound and one trough
is locally stable for @xmath170 .
since structures with several mounds and troughs approach this steady - state structure through the coarsening process , it is reasonable to expect that fixed points of the noiseless equation with more than one mounds and troughs would be unstable .
we have numerically obtained fixed points of eq.([cld ] ) with two mounds and troughs for different values of the sample size and the separation between the peaks of the two mounds .
the slope of the mounds at these fixed points is found to be the same as that in the fixed point with one mound and one trough .
we find that the stability matrix for such fixed points always has a real , positive eigenvalue , indicating that the structure is unstable and would evolve to the stable configuration with one mound and one trough .
the magnitude of the positive eigenvalue of the stability matrix for two - mounded fixed points depends on the sample size , the separation between the peaks of the mounds and the relative heights of the mounds in a complicated way .
we have not been able to extract any systematic quantitative information from these dependences .
we find a qualitative trend indicating that the magnitude of the positive eigenvalue decreases as the separation between the peaks of the two mounds is increased .
since the time scale of the development of the instability of two - mounded structures is given by the inverse of the positive eigenvalue , this result is consistent with the expectation that the time required for two mounds to coalesce should increase with the separation between the mounds .
these results suggest that the coarsening of the mounds in model i reflects the instability of structures with multiple mounds and troughs .
if this is true , then coarsening of mounds should be observed in this model even when the noise term in eq.([cld ] ) is absent . to check this ,
we have carried out numerical studies of coarsening in the noiseless version of eq .
( [ cld ] ) . in these studies ,
the time evolution of an initial configuration with a pillar of height @xmath197 at the central site of an otherwise flat interface is followed numerically in the presence of noise until the instability caused by the presence of the pillar is well developed .
the profiles obtained for different realizations of the noise used in the initial time evolution are then used as initial configurations for coarsening runs without noise . the dotted line in fig.[bigfig18 ]
shows the width versus time data obtained from this calculation .
the coarsening exponent in the absence of noise is found to be the same ( @xmath198 ) as that of the noisy system , indicating that the coarsening in this model is driven by processes associated with the deterministic part of the growth equation .
we have examined the details of the process by which two mounds coalesce to form a single one .
the different steps in this process are illustrated in the snapshots of interface profiles shown in fig.[bigfig17 ] where one can see how the two mounds near the center come together to form a single one as time progresses .
first , the separation between the peaks of the mounds decreases with time . when this distance becomes sufficiently small , the `` v''-shaped segment that separates the peaks of the mounds `` melts '' to form a rough region with many spikes .
this region of the interface then self - organizes to become the top part of a mound .
although the data shown in this figure were obtained for model ii , it also represents quite closely the process of coalescence of mounds in models i and ia .
an estimate of the time scale associated with the second part of this process , during which the `` melted '' region of the interface transforms into the top part of a mound , may be obtained in the following way .
the absolute value of the closest - to - zero eigenvalue of the stability matrix for the single - mounded fixed point provides an estimate of the time scale over which configurations close to the fixed point evolve to the fixed point itself . in the inset of fig.[bigfig18 ] , we have shown the dependence of the magnitude @xmath199 of the closest - to - zero eigenvalue for @xmath87 , @xmath88 on the system size @xmath70 .
the eigenvalue scales with the system size as @xmath200 , indicating that the time scale for the decay of fluctuations with length scale @xmath70 is proportional to @xmath201 .
this is consistent with the observed value of the coarsening exponent , @xmath198 , which indicates that the time scale @xmath202 for the coalescence of mounds separated by distance @xmath67 is proportional to @xmath203 .
we have found very similar behavior for the closest - to - zero eigenvalue of the stability matrix for the single - mounded fixed point of model ia in which a different form of the control function is used .
coarsening data for this model are shown in fig.[bigfig18 ] . in this model , there is a long time interval between the onset of the instability and the beginning of power - law coarsening . during this time interval ,
the interface segments near the tall pillars formed at the instability organize themselves into triangular mounds .
this process produces a plateau in the width versus time plot .
eventually , however , power - law coarsening with @xmath204 is recovered , as shown by the dashed line in fig.[bigfig18 ] .
since the onset of power - law coarsening in this model occurs at very late times , we could not get coarsening data for this model over a very wide time interval .
consequently , the calculated value of the coarsening exponent for this model is less accurate .
however , our results show quite clearly that the `` universal '' features of the coarsening dynamics of models i and ia are the same . going back to the discrete model ii
, we first examined its coarsening dynamics in the absence of noise .
the noiseless limit of this model is not well - defined in the sense that there is no explicit noise term that can be turned off .
the stochasticity in this model arises from two sources : first , the randomness associated with the selection of the deposition site @xmath42 ( the quantity @xmath205 defined in eq.([model2 ] ) is calculated at this site and at its nearest - neighbor sites ) ; and second , the randomness in the selection of one of the two neighbors of site @xmath42 in case of a tie in their values of @xmath60 . in order to make the dynamics deterministic
, we employ a parallel update scheme in which all the lattice sites , @xmath206 , are updated simultaneously instead of sequentially .
this eliminates the stochasticity arising from the choice of the sequence in which the sites are updated .
the randomness associated with the selection of a neighbor in case of a tie is eliminated by choosing the right neighbor if the serial number of the occurrence of a tie , measured from the beginning of the simulation , is even , and the left neighbor if the serial number is odd . with these modifications of the update rules
, the system evolves in a perfectly deterministic way . to study coarsening in this deterministic version of model ii , we prepare an initial structure with two identical mounds separated by distance @xmath207 . the slope of these mounds is chosen to be equal to the `` selected '' value found in simulations of the original model .
we then study the time evolution of this structure according to the parallel dynamics defined above , monitoring how the distance @xmath67 between the peaks of the mounds changes with time @xmath32 .
we find that the the value of @xmath67 increases initially , in order to accommodate a slightly higher value of the selected slope for the parallel dynamics . after reaching a maximum value that is @xmath208 higher than @xmath207 , the distance @xmath67 decreases with time , indicating that the noiseless dynamics leads to coarsening . eventually , the two mounds coalesce into a single one and the system remains in the state with one mound and one trough at later times .
the slope of the mounds remains constant during the coarsening process .
assuming power - law coarsening with exponent @xmath9 , the separation @xmath67 at time @xmath32 is expected to have the form @xmath209 where @xmath210 is a constant .
this form implies that the time @xmath211 required for the coalescence of two mounds separated by distance @xmath207 is proportional to @xmath212 . in fig.[bigfig20 ] , we have shown the time dependence of @xmath213 for three different initial separations , @xmath214 and 100 , and fits of the data to the form of eq.([xoft ] ) , yielding the result @xmath215 .
only the data for @xmath67 at times larger than the time at which it returns to @xmath207 after the initial increase are shown in the figure . from these observations , we conclude that the coarsening exponent in the zero - noise limit of model ii is the same ( @xmath216 ) as that found for the two versions of the continuum model .
the observation that the coarsening exponent for the noisy version of model ii is different from 1/3 then indicates that the effect of noise in the discrete model is _ qualitatively different _ from that in the continuum models .
we discuss below a possible explanation of this behavior .
the fact that noiseless versions of all three models exhibit the same value of the coarsening exponent ( @xmath217 ) suggests that the coarsening is driven by an effective attractive interaction between the peaks of neighboring mounds .
the observed value of @xmath9 suggests @xcite that the this attractive interaction is proportional to @xmath218 where @xmath67 is the separation between the mound tips .
this interaction would lead to the observed result , @xmath219 , in the noiseless limit if the rate of change of @xmath67 with @xmath32 is assumed to be proportional to the attractive force ( `` overdamped limit '' ) .
the presence of noise in the original growth model leads to a noise term in the equation of motion of the variable @xmath67 , but the nature of this noise term is not clear . since the observed coarsening dynamics in the noisy model ii ( @xmath220 ) suggests a similarity with random walks , we propose that the effective dynamics of the variable @xmath67 is governed by the kinetic equation @xmath221 where @xmath25 is a gaussian , delta - correlated noise with zero mean and variance equal to @xmath222 . in this phenomenological description ,
the coarsening of a two - mounded structure in model ii is mapped to a brownian walk of a particle in an attractive potential field with an absorbing wall at the origin , such that the particle can not escape once it arrives at the origin .
the absorption of a particle at the origin corresponds to the coalescence of two mounds in the original height picture .
thus , in this reduced model , the quantity of interest is the typical first passage time @xmath223 ( i.e. the time taken by a particle to reach the origin ) as a function of @xmath207 , the initial distance of the particle from the origin . in the noiseless limit ( @xmath224 )
, @xmath223 is equal to @xmath225 , and in the purely brownian walk limit ( @xmath226 ) , the typical value of @xmath223 should be of the order of @xmath227 .
therefore , for sufficiently large values of @xmath207 , random - walk behavior characterized by @xmath228 is expected .
however , for relatively small values of @xmath207 , the behavior should be dominated by the attractive interaction .
therefore , we expect that the dynamics described by eq.([bw ] ) with nonzero @xmath210 and @xmath229 would exhibit a crossover from a noise dominated regime to an interaction dominated regime as the value of @xmath207 is decreased . this crossover is expected to occur near @xmath230 , for which the values of @xmath223 obtained from the two individual terms in eq.([bw ] ) become comparable . we , therefore , propose a scaling form for the dependence of @xmath223 on @xmath207 : @xmath231 where the scaling function @xmath232 has the following asymptotic dependence on its argument @xmath11 : @xmath233 for @xmath234 and @xmath235 for @xmath236 .
our numerical study of the reduced model of eq.([bw ] ) confirms the validity of this scaling ansatz .
note that for any nonzero value of @xmath229 , @xmath211 would be proportional to @xmath237 , implying @xmath228 , for sufficiently large values of @xmath207 . to test the validity of this reduced description of the coarsening dynamics of the original model ii with stochastic sequential updates
, we have simulated the evolution of a two - mounded structure in this model .
the two - mounded structure used in these simulations is identical to that used in the study of coarsening in the model with parallel updates . in these simulations
also , the average value of @xmath67 exhibits a small initial increase followed by a steady decrease .
the initial growth of @xmath238 with time is found to be linear , indicating the presence of a random additive noise in the effective equation of motion of @xmath67 .
fig.[bigfig21 ] shows the time dependence of @xmath239 obtained from simulations of @xmath93 samples of model ii with @xmath94 , @xmath95 , and @xmath240 . in this plot ,
the origin of time has been shifted to the point where @xmath241 returns to the initial value @xmath240 after the small initial increase , and @xmath242 units of simulation `` time '' ( number of deposited layers ) is taken to be the unit of @xmath32 .
the number of independent runs used in the calculation of averages is 800 .
the observed dependence of @xmath239 on @xmath32 can be described reasonably well by the reduced equation of eq.([bw ] ) for appropriate choice of the values of the parameters @xmath210 and @xmath229 . as shown in fig.[bigfig21 ] ,
the @xmath239 calculated numerically from eq.([bw ] ) with @xmath243 and @xmath244 provides a good fit to the data obtained from simulations of the growth model . due to the limited range of the simulation data
, the values of @xmath210 and @xmath229 can not be determined very accurately from such fits : values of @xmath210 in the range @xmath245 and values of @xmath229 in the range @xmath246 ( larger values of @xmath210 have to be combined with smaller values of @xmath229 ) provide reasonable descriptions of the simulation data . for such values of @xmath210 and @xmath229 , and @xmath247 where @xmath248 is the sample size used in the calculation of the coarsening exponent ,
@xmath249 is of order unity , indicating that the effects of the noise term in eq.([bw ] ) should be observed in the simulation data .
we , therefore , conclude that the presence of an additive random noise term in the effective equation of motion for the separation between the peaks of neighboring mounds in model ii is a plausible explanation for the observed value of the coarsening exponent , @xmath250 . in view of this conclusion , it is interesting to enquire why the coarsening exponent @xmath9 for models i and ia has the value @xmath251 characteristic of dynamics governed by the deterministic interaction between mound tips .
we can not provide a conclusive answer to this question .
one possibility is that the additive random noise in the original growth equations for these models does not translate into a similar noise in the effective equation of motion for the separation of mound tips .
a second possibility is that the equation of motion for the separation @xmath67 for these models also has the form of eq.([bw ] ) , but the relative strength of the noise is much smaller , so that the crossover value @xmath252 is much larger than the typical sample sizes used in our simulations . under these circumstances , the dynamics of @xmath67 would be governed by the interaction and @xmath223 would be proportional to @xmath253 , giving the value @xmath251 for the coarsening exponent @xmath9 .
if the second explanation is correct , then one should observe a crossover from @xmath216 to @xmath228 in models i and ia as the sample size @xmath70 is increased .
we do not find any evidence for such a crossover in our simulations .
in passing , we note that the dynamics of the slope variables @xmath254 is strictly conserved in the models studied here if periodic or fixed boundary conditions are used .
also , the deterministic part of the growth equations conserves the integrated height .
the `` magnetization '' of the ising variables @xmath180 representing the signs of @xmath254 is not strictly conserved : it is conserved only in a statistical sense . however , unlike other well - known examples @xcite of systems with conserved dynamics , we obtain in model ii a coarsening exponent that is different from the expected lifshitz - slyozov value , @xmath216 . since the height variables in model ii are integers , there are some sites for which @xmath255 .
the assignment of the value of the ising variable @xmath179 at such sites is clearly ambiguous
. it may be more appropriate to use a three - state variable , taking the values @xmath256 and @xmath257 , to describe the coarsening behavior of this model .
this problem does not arise in models i and ia for which the height variables are real numbers because the likelihood of two neighboring height variables to be exactly the same is vanishingly small .
as discussed in section [ pt ] , the properties of the mounded phase of model i are determined to a large extent by the mounded fixed point of the deterministic part of the equations of motion of the height variables .
the presence of noise changes the critical value of the control parameter @xmath66 from @xmath129 to @xmath258 , but does not affect strongly the properties of the mounded steady state of the system ( see , for example , fig.[bigfig11 ] ) .
therefore , we expect that the properties of the mounded phase would not change drastically if the statistics of the noise is altered . on the other hand , it is well - known @xcite that the exponents that describe the scaling behavior in the kinetically rough phase depend strongly on the nature of the noise .
in particular , the exponents for conserved noise are expected to be quite different from those describing the behavior for nonconserved noise .
the occurrence of the nonlinear instability that leads to the mounded phase in our models is contingent upon the spontaneous formation of pillars of height @xmath197 , if the initial state of the system is completely flat .
the probability of formation of such pillars depends crucially @xcite on the values of the roughening exponents which , in turn , are strongly dependent on the nature of the noise .
therefore , we expect that the nature of the noise may be very important in determining whether the instability leading to mound formation actually occurs in samples with flat initial states .
we have investigated this issue in detail by carrying out simulations of a version of model i in which the noise is conserved @xcite .
the equations of motion for the height variables in this model have the form of eq.([cld ] ) , with the noise terms @xmath259 having the properties @xmath260 where @xmath261 if @xmath262 and zero otherwise .
this model is expected to exhibit kinetic roughening with exponents @xmath263 , @xmath264 , and @xmath265 in one dimension @xcite .
since the value of @xmath2 for this model is less than unity , it exhibits conventional dynamical scaling with the typical value of the nearest - neighbor height difference saturating at a constant value at long times .
the value of this constant is expected to increase @xcite as the strength @xmath78 of the nonlinearity is increased . as discussed in detail in ref .
, the nonlinear instability that leads to mound formation is expected to occur in the time evolution of such models from a perfectly flat initial state only if the value of @xmath78 is sufficiently large to allow the spontaneous , noise - induced formation of pillars of height greater than @xmath266 .
so , if the value of @xmath78 is sufficiently small , then the model with conserved noise , evolving from a flat initial state , would not exhibit the mounding transition . on the other hand , if the instability is induced in the model by starting the time evolution from a state in which there is at least one pillar with height greater than @xmath83 , then it is expected to evolve to the mounded state if the value of @xmath66 is sufficiently small to make the mounded state stable .
so , the long - time steady state of the conserved model is expected to exhibit an interesting dependence on the initial state : if @xmath78 is sufficiently small ( so that pillars with height greater than @xmath83 are not spontaneously generated in the time evolution of the interface from a flat initial state ) , and the value of @xmath66 sufficiently small ( so that the mounded state is stable in the presence of noise ) , then the steady state would be kinetically rough if the initial state is sufficiently smooth , and mounded if the initial state contains pillars of height greater than @xmath83 .
this `` bistability '' does not occur for the nonconserved model i because the nearest - neighbor height difference in this model continues to increase with time , so that the instability always occurs at sufficiently long times @xcite .
our simulations of the model with conserved noise show the bistable behavior discussed above in a large region of the @xmath139 plane .
we find that in this model , the height of a pillar on an otherwise flat interface increases in time if its initial value @xmath81 is larger than @xmath267 ( the dependence of @xmath83 on @xmath66 is weak ) .
this dependence of @xmath83 on @xmath78 is very similar to that @xcite found for model i with nonconserved noise .
we also find that the typical values of the nearest - neighbor height difference do not continue to increase with time in this model .
consequently , if @xmath78 is sufficiently small , pillars with height greater than @xmath83 are not generated , and the system exhibits conventional kinetic roughening with exponent values close to the expected ones @xcite . on the other hand ,
if the time evolution of the same system is started from a state with a pillar of height greater than @xmath83 , then it evolves to a mounded state very similar to the one found in the model with nonconserved noise if the value of @xmath66 is sufficiently small .
the two steady states obtained for the conserved model with @xmath87 , @xmath121 are shown in fig.[bigfig22 ] .
the long - time state obtained in a run starting from a flat configuration is kinetically rough , whereas the state obtained in a run in which the nonlinear instability is initially seeded in the form of a single pillar of height @xmath268 at the central site is mounded with a well - defined slope , as in model i with nonconserved noise .
the difference between the two profiles , obtained for the same parameter values for two different initial states , is quite striking .
since the steady state in the conserved model depends on the initial condition , it is not possible to draw a conventional phase diagram for this model in the @xmath269 plane : the transition lines are different for different initial conditions . in fig.[bigfig23 ] , we have shown three transition lines for this model in the @xmath139 plane .
the line drawn through the circles ( line 1 ) is obtained from simulations in which the system is started from a flat initial state . if @xmath78 is small , then the steady state reached in such runs is kinetically rough for all values of @xmath66 . as @xmath78 is increased above a `` critical value '' @xmath270 , pillars with height greater than @xmath83 are spontaneously generated during the time evolution of the system and it exhibits a transition to the mounded state if the value of @xmath66 is not very large .
the circles represent the values of @xmath66 for which 50% of the runs show transitions to the mounded state . the line through the diamonds ( line 2 ) corresponds to 50% probability of transition to the mounded state from an initial state with a pillar of height @xmath271 1000 on an otherwise flat interface
. the probability of reaching a mounded steady state in such runs decreases from unity as the value of @xmath66 is increased , and falls below 50% as line 2 is crossed from below . for large @xmath78 ,
lines 1 and 2 merge together .
this is expected : the probability of occurrence of a mounded steady state should not depend on how the pillars that initiate the nonlinear instability are generated .
the third line ( the one passing through the squares ) represents 50% probability of transition to the kinetically rough state from a mounded initial state ( the fixed point of the noiseless equations of motion with one mound and one trough ) .
this line reflects the noise - induced instability of the mounded steady state for relatively large values of @xmath66 .
the differences between lines 2 and 3 are due to finite - time hysteresis effects similar to those discussed in section [ pt ] in the context of determining the critical value @xmath142 of the control parameter @xmath66 for model i with nonconserved noise .
the interesting region in the `` phase diagram '' of fig.[bigfig23 ] is the area enclosed by lines 1 and 2 and the @xmath272 line . for parameter values in this region , the system exhibits bistable behavior , as discussed above .
this bistability is unexpected in the sense that in most studies of nonequilibrium surface growth , it is implicitly assumed that the long - time steady state of the system does not depend on the choice of the initial state .
so , it is important to examine whether the dependence of the steady state on the initial condition in the conserved model reflects a very long ( but finite ) transition time from one of the two apparent steady states to the other one .
we have addressed this question by carrying out long ( @xmath32 of the order of @xmath190 ) simulations of small samples with parameter values in the middle of the `` bistable region '' for flat and mounded initial states .
we did not find any evidence for transitions from one steady state to the other one in such simulations .
of course , we can not rule out the possibility that such transitions would occur over much longer time scales .
to summarize , we have shown from numerical simulations that a class of 1d surface growth models exhibits mound formation and power - law coarsening of mounds with slope selection as a result of a nonlinear instability that is controlled by the introduction of an infinite series of terms with higher - order gradient nonlinearities .
the models considered here are discretized versions of a well - known continuum growth equation and an atomistic model originally formulated for providing a discrete realization of the growth equation .
we have shown that these models exhibit a dynamical phase transition between a kinetically rough phase and a mounded phase as a parameter that measures the effectiveness of controlling the instability is varied .
we have defined an order parameter for this first - order transition and used finite - size scaling to demonstrate how the sample - size dependence of this order parameter provides a clear distinction between the rough and mounded phases .
we have also mapped out the phase boundary that separates the two phases in a two - dimensional parameter space .
we would like to emphasize that the es mechanism , commonly believed to be responsible for mound formation in surface growth , is not present in our models .
our models exhibit a nonlinear instability , instead of the linear instability used conventionally to represent the effects of es barriers .
the mechanism of mound formation in our models is also different from a recently discovered @xcite one involving fast edge diffusion , which occurs in two or higher dimensions .
the slope selection found in our models is a rare example of pattern formation from a nonlinear instability .
this is clearly different from slope selection in es - type models in which the mounds maintain a constant slope during coarsening only if the nonequilibrium surface current vanishes at a particular value of the slope .
the selected slope in such models is simply the slope at which the current is zero .
the behavior of our models is more complex : in these models , the surface current is zero for all values of constant slope , and the selected value of the slope is obtained from a nonlinear mechanism of pattern selection .
our studies bring out two other unexpected results .
we find that the coarsening behavior of an atomistic model ( model ii ) specifically designed to provide a discrete realization of the growth equation that leads to model i is different from that of model i : the exponents that describe the power - law coarsening are different in the two models .
we show that this difference may arise from a difference in the nature of the effective noise that enters the equation of motion for the separation between neighboring mounds in the two cases .
the second surprising result is that the numerically obtained long - time behavior of model i with conserved noise in a region of parameter space depends crucially on the initial conditions : the system reaches a mounded or kinetically rough steady state depending on whether or not the initial state is sufficiently rough . to our knowledge , this is the first example of `` nonergodic '' behavior in nonequilibrium surface growth .
the behavior found in our 1d models may be relevant to experimental studies of the roughening of steps on a vicinal surface .
as noted earlier , the form of the control function @xmath56 used in model ia is physically reasonable .
however , since very little is known about the values of the model parameters appropriate for experimentally studied systems , we are unable to determine whether the mechanism of mound formation found in our study would be operative under experimentally realizable conditions .
the nonlinear instability found in our 1d models is also present @xcite in the experimentally interesting two - dimensional version of the growth equation of eq.([lds1 ] ) .
however , it is not clear whether this instability , when controlled in a manner similar to that in our 1d models , would lead to the formation of mounds in two dimensions .
this question is currently under investigation .
since the growth equation of eq.([lds1 ] ) exhibits conventional dynamic scaling in the kinetic roughening regime in two dimensions , the nonlinear instability would not occur in runs with flat initial states if the value of @xmath78 is small .
therefore , the behavior in two dimensions is expected to be similar to that of our 1d model i with conserved noise : the nature of the long - time steady state may depend crucially on initial conditions in a region of parameter space . such nonergodic behavior , if found in two dimensions , would have interesting implications for the growth of films on patterned substrates .
all the results described in this paper have been obtained from numerical studies of models that are discrete in both space and time .
it is interesting to enquire whether the truly continuum growth equation of eq.([lds1 ] ) exhibits similar behavior .
this question acquires special significance in view of studies @xcite that have shown that discretization may drastically change the behavior of nonlinear growth equations similar to eq.([lds1 ] ) . since the interesting behavior found in our discretized models arises from the nonlinear instability found earlier @xcite ,
the question that we have to address is whether a similar instability is present in the truly continuum growth equation .
this question was addressed in some detail in ref.@xcite where it was shown that the nonlinear instability is not an artifact of discretization of time or the use of the simple euler scheme for integrating the the growth equation . in the present study , we have found additional evidence ( see section [ mound ] ) that supports this claim . we should also point out that the atomistic model ii , for which the question of inaccuracies arising from time integration does not arise , exhibits very similar behavior , suggesting that the behavior found in models i and ia is not an artifact of the time discretization used in the numerical integration . the occurrence of the nonlinear instability does depend on the way space is discretized ( i.e. how the lattice derivatives are defined ) . in earlier work @xcite as well as in the present study ,
the lattice derivatives are defined in a left - right symmetric way .
we have found that the instability actually becomes stronger if the number of neighbors used in the definition of the lattice derivatives is increased .
this result suggests that the instability is also present in the continuum equation .
it has been found by putkaradze _
et al . _
@xcite that the instability does not occur if the lattice derivatives are defined in a different way in which either left- or right - discretization of the nonlinear term is used , depending on the sign of the local slope of the interface .
there is no reason to believe that this definition is `` better '' or more physical than the symmetric definitions used in our work .
the only rigorous result we are aware of for the behavior of eq.([lds1 ] ) is derived in ref.@xcite where it is shown that the solutions of the noiseless equation are bounded for sufficiently smooth initial conditions .
this result , however , does not answer the question of whether the instability occurs in the continuum equation . as discussed in ref .
@xcite , the nonlinear instability of eq.([lds2 ] ) , signalled by a rapid initial growth of the height ( depth ) of isolated pillars ( grooves ) , may not lead to a true divergence of the height variables .
the results reported in the present work would remain valid as long as high pillars or deep grooves are formed by the instability the occurrence of a true divergence is not necessary . in the present work , we have shown ( see section [ pt ] ) that the continuum equation with @xmath56 defined in eq.([pvform ] ) does admit stationary solutions that exhibit all the relevant features of stationary solutions of the discretized equation .
this result provides additional support to the contention that the behaviors of the continuum and discretized systems are qualitatively similar .
we should , however , mention that these stationary solutions of the continuum equation do not pick out a selected slope of the interface : profiles similar to those shown in fig.[bigfig11 ] may be obtained for different values of the parameter @xmath117 in eq.([cont ] ) that determines the slope of the triangular mound .
slope selection in the continuum equation may occur as a consequence of the requirement of local stability of such stationary solutions .
as mentioned in section [ pt ] , we have not attempted a linear stability analysis of such numerically obtained stationary solutions of the continuum equation because doing such a calculation without discretizing space would be extremely difficult .
further investigation of this question would be useful .
finally , we would like to emphasize that the discrete models studied here would continue to be valid models for describing nonequilibrium surface growth even if the behavior of the truly continuum growth equation of eq.([lds1 ] ) turns out to be different from that found here .
these models may be looked upon as ones in which continuous ( in models i and ia ) or discrete ( in model ii ) height variables defined on a discrete lattice evolve in continuous or discrete time .
these models have all the correct symmetries and conservation laws of the physical problem , and they exhibit , for different values of the control parameter @xmath66 , both the phenomena of kinetic roughening and mound formation found in experiments .
there is no compelling reason to consider the continuum equation to be more `` correct '' or `` physical '' than these models .
epitaxial growth is intrinsically a discrete process at the molecular level and a continuum description is an approximation that may not be valid in some situations . from a different perspective
, the nonequilibrium first - order phase transition found in our models is interesting , especially because it occurs in 1d systems with short range interactions .
such phase transitions have been found earlier in several 1d `` particle hopping '' models @xcite .
it would be interesting to explore possible connections between such models and the 1d growth models studied here .
we thank serc , iisc for computational facilities , and s. das sarma and s. s. ghosh for useful discussions .
currently at the department of physics , condensed matter theory center , university of maryland , college park , md 20742 .
barabasi and h. e. stanley , _ fractal concepts in surface growth _
( cambridge university press , cambridge , 1995 ) .
j. krug , adv .
phys . * 46 * , 139 ( 1997 ) .
j. krim and g. palasantzas , int .
j. mod . phys . * b9 * , 599 ( 1997 ) .
m. d. johnson , c. orme , a. w. hunt , d. graff , j. sudijono , l. m. sander and b. g. orr , phys .
lett * 72 * , 116 ( 1994 ) .
k. thurmer , r. koch , m. weber and k. h. rieder , phys .
lett . * 75 * , 1767 ( 1995 ) .
j. a. stroscio , d. t. pierce , m. stiles , a. zangwill and l. m. sander , phys . rev .
lett * 75 * , 4246 ( 1995 ) .
f. tsui , j. wellman , c. uher and r. clarke , phys .
lett . * 76 * , 3164 ( 1996 ) .
k . zuo and j. f. wendelken , phys .
78 * , 2791 ( 1997 ) .
m. f. gyure , j. k. zinck , c. ratsh and d. d. vvendsky , phys .
lett . * 81 * , 4931 ( 1998 ) .
g. apostolopoulos , j. herfort , l. daweritz and k. h. ploog , phys .
. lett . * 84 * , 3358 ( 2000 ) . y .-
zhao , h .-
yang , g .- c .
wang and t .- m .
lu , phys .
b * 57 * , 1922 ( 1998 ) .
g. lengel , r. j. phaneuf , e. d. williams , s. das sarma , w. beard and f. g. johnson , phys .
b * 60 * , r8469 ( 1999 ) .
m. siegert and m. plischke , phys .
lett . * 73 * , 1517 ( 1994 ) .
m. rost and j. krug , phys .
e * 55 * , 3952 ( 1997 ) .
m. siegert , phys .
lett . * 81 * , 5481 ( 1998 ) .
l. golubovic , phys .
lett . * 78 * , 90 ( 1997 ) . o. pierre - louis , m. r. dorsogna and t. l .
einstein , phys .
lett . * 82 * , 3661 ( 1999 ) .
m. v. ramana murty and b. h. cooper , phys .
lett . * 83 * , 352 ( 1999 ) .
s. das sarma and p. punyindu , surf .
lett . * 424 * , l339 ( 1998 ) .
m. biehl , w. kinzel , and s. schinzer , europhys .
* 41 * , 443 ( 1998 ) .
j. krug , j. stat .
phys . * 87 * , 505 ( 1997 ) .
i. elkinani and j. villain , j. phys .
i * 4 * , 949 ( 1994 ) .
p. politi and j. villain , phys .
b * 54 * , 5114 ( 1996 ) .
s. das sarma , p. punyindu and z. toroczkai , surf .
lett . * 457 * , l309 ( 2000 ) .
p. punyindu - chatraphorn , z. toroczkai and s. das sarma , phys .
b * 64 * , 205407 ( 2001 ) .
s. das sarma , s. v. ghaisas , and j. m. kim , phys .
e * 49 * , 122 ( 1994 ) .
g. ehrlich and f. g. hudda , j. chem . phys . *
44 * , 1039 ( 1996 ) ; s. c. wang and g. ehrlich , phys .
lett . * 70 * , 41 ( 1993 ) .
r. l. schwoebel and e. j. shipsey , j. appl . phys . * 37 * , 3682 ( 1966 ) ; r. l. schwoebel , j. appl . phys . * 40 * , 614 ( 1969 ) . j. villain , j. phys .
i ( france ) * 1 * , 19 ( 1991 ) .
z. w. lai and s. das sarma , phys .
lett . * 66 * , 2348 ( 1991 ) . j. m. kim and s. das sarma , phys .
lett . * 72 * , 2903 ( 1994 ) . c. dasgupta , s. das sarma , and j. m. kim , phys .
e * 54 * , r4552 ( 1996 ) . c. dasgupta , j. m. kim , m. dutta , and s. das sarma , phys .
e * 55 * , 2235 ( 1997 ) .
a. kundagrami and c. dasgupta , physica a * 270 * , 135 ( 1999 ) . t. sun , h. guo , and m. grant , phys . rev .
a * 40 * , 6763 ( 1989 ) . b. chakrabarti and c. dasgupta , europhys .
( in press )
. l. f. shampine , _ numerical solution of ordinary differential equations _
( chapman & hall , new york , 1994 ) .
z. racz , m. siegert , d. liu and m. plischke , phys .
a * 43 * , 5275 ( 1991 ) .
ma , _ modern theory of critical phenomena _
( benjamin , reading , 1976 ) .
z. toroczkai , g. corniss , s. das sarma and r. k. p. zia , phys .
e. * 62 * , 276 ( 2000 )
. m. r. evans , brazilian j. phys . *
30 * , 42 ( 2000 ) [ cond - mat/0007293 ] .
a. j. bray , adv .
phys * 43 * , 357 ( 1994 ) .
a. d. rutenberg and a. j. bray , phys .
e * 50 * , 1900 ( 1994 ) .
t. newmann , and a. j. bray , j. phys .
a * 29 * , 7917 ( 1996 ) .
v. putkaradze , t. bohr and j. krug , nonlinearity * 10 * , 823 ( 1997 ) . | we study a class of one - dimensional , nonequilibrium , conserved growth equations for both nonconserved and conserved noise statistics using numerical integration .
an atomistic version of these growth equations is also studied using stochastic simulation .
the models with nonconserved noise statistics are found to exhibit mound formation and power - law coarsening with slope selection for a range of values of the model parameters . unlike previously proposed models of mound formation , the ehrlich - schwoebel step - edge barrier , usually modeled as a linear instability in growth equations , is absent in our models .
mound formation in our models occurs due to a nonlinear instability in which the height ( depth ) of spontaneously generated pillars ( grooves ) increases rapidly if the initial height ( depth ) is sufficiently large .
when this instability is controlled by the introduction of an infinite number of higher - order gradient nonlinearities , the system exhibits a first - order dynamical phase transition from a rough self - affine phase to a mounded one as the value of the parameter that measures the effectiveness of control is decreased .
we define a new `` order parameter '' that may be used to distinguish between these two phases . in the mounded phase ,
the system exhibits power - law coarsening of the mounds in which a selected slope is retained at all times .
the coarsening exponents for the continuum equation and the discrete model are found to be different .
an explanation of this difference is proposed and verified by simulations . in the growth equation with conserved noise , we find the curious result that the kinetically rough and mounded phases are both locally stable in a region of parameter space
. in this region , the initial configuration of the system determines its steady - state behavior . 2 |
the financial crisis has boosted the development of several network - based methodologies to monitor systemic risk in the financial system @xcite .
a traditional approach towards the quantification of systemic risk is to measure the effects of a shock on the external assets of each institution and then to aggregate the losses .
however , the crisis has highlighted that stress - testing should also incorporate so called `` second - round '' effect , which might arise via interbank exposures , either as losses on the asset side or liquidity shortages ( see e.g. @xcite and references therein ) .
for instance , the recent ecb comprehensive assessment carried out in 2014 @xcite goes into this direction by taking into account counterparty credit risk while the basel iii framework @xcite gives attention to interconnectedness as a key source of systemic risk .
some network - based methods focus on the events of a bank s default ( i.e. its equity going to zero ) as the only relevant trigger for the contagion to be passed on to the counterparties . in other words ,
an institution that has faced some shocks will not affect its counterparties in any way as long as it is left with some positive equity .
this is a useful simplification which has allowed for a number of mathematical developments @xcite . because regulators recommend banks to keep their largest single exposure well below their level of equity , most stress test conducted in this way yield essentially to the result that a single initial bank default never triggers any other default .
systemic risk emerges only if , at the same time , one assumes a scenario of weak balance sheets @xcite or a scenario of fire sales @xcite .
in contrast , both the intuition and the classic merton approach , suggest that the loss of equity of an institution , even with no default , will imply a decrease in the market value of its obligations to other institutions . in turn , this means a loss of equity for those institutions , as long as they revalue their equity as the difference between assets and liabilities .
therefore , financial distress , meant as loss of equity , can spread from a bank to another although no default occurs in between .
the total loss of equity in the system can be substantial even if no bank ever defaults in the process .
indeed , in the 2007/2008 crisis , losses due to the mark - to - market re - evaluation of counterparty risk were much higher than losses due from direct defaults . ]
the so - called debtrank methodology has been developed with the very idea to capture such a distress propagation @xcite .
the impact of a shock , as measured by debtrank , is fully comparable to the traditional default - only propagation mechanisms @xcite in the sense that the latter is a lower bound for the former . in other words , debtrank measures at least the impact that one would have with the defaults - only , but it is typically larger and this allows to assign a level of systemic importance in most situations in which the traditional method would be unable to do so because the impact would be zero for all banks .
debtrank has been applied to several empirical contexts @xcite but it was not so far been embedded into a stress - test framework . in this paper , building on the method introduced in @xcite , we develop a stress - test _ framework _ aimed at providing central bankers and practitioners with a monitoring tool of the network effects .
the main contributions of our works are as follows .
first , the framework delivers not only an estimation of _ first - round _ ( shock on external assets ) , and _ second - round _ ( distress induced in the interbank network ) effects , but also a _ third - round _ effect consisting in possible further losses induced by fire sales . to this end
we incorporate a simple mechanism by which banks determine the necessary sales of the asset that was shocked in order to recover their previous leverage level and assuming a linear market impact of the sale on the price of the asset .
the three effects are disentangled and can be tracked separately to assess their relative magnitude according to a variety of scenarios on the initial shock on external assets and on liquidity of the asset market .
second , the framework allows to monitor at the same time the _ impact _ and the _ vulnerability _ of financial institutions . in other words , institutions
whose default would cause a large loss to the system become problematic only if they are exposed to large losses when their counterparties or their assets get shocked .
these quantities are computed through two _ networks of leverage _ that are the main linkage between the notion of capital requirements and the notion of interconnectedness .
third , the framework allows to estimate loss distributions both at the individual bank level and at the global level , allowing for the computation of individual and global var and cvar ( table [ tab : variables ] ) .
fourth , since data on bilateral exposures are seldom available , the framework includes a module to estimate the interbank network of bilateral exposures given the information on the total lending and borrowing of each bank . here
, we use a combination of fitness model @xcite , for the network structure and an iterative fitting method to estimate the lending volumes , but alternative methods could be used or added as benchmark comparison ( e.g. the maximum entropy method @xcite , or the minimum density method @xcite .
finally , the framework has been developed in matlab and is available upon request to the authors .
as an illustration , we carry out a stress - test exercise on a dataset of 183 european banks over the years @xmath0@xmath1 , starting from the estimation of their interbank exposures .
this paper is organized as follows . in section [ sec : related work ] we review similar or related work ; in section [ sec : stress - test ] we describe the main aspects of the framework , providing an outline of the distress process , a discussion of the main variables , and the framework s building blocks ; in section [ sec : exercise ] , we show how the framework can be applied to a dataset and we discuss the main results of this exercise ; in section [ sec : discussion ] , we review the main contributions and introduce elements for future research . in appendix
[ sec : methods ] , we provide the technical details of the distress propagation process , including how the key measures are computed ; in the appendix [ sec : data_collection ] , we described the data we used for the exercise in section [ sec : exercise ] and , last , in appendix [ sec : network_reconstruction ] , we outline the network reconstruction methods when only the total interbank lending / borrowing for each bank is known .
the recent and still ongoing economic and financial crisis has made clear the importance of methods of early detection of systemic risk in the financial system . in particular , researchers , regulators and policy - makers have recognized the importance of adopting a macroprudential approach to understand and mitigate financial stability .
notwithstanding the many efforts @xcite , regulators still lack an adequate framework to measure and address systemic risk .
the traditional micro - prudential approach consists in trying and ensuring the stability of the banks , one by one , with the assumption that as long as each unit is safe the system is safe .
this approach has demonstrated to be a dangerous over - simplification of the situation @xcite .
indeed , we have learned that it is precisely the interdependence among institutions , both in terms of liabilities or complex financial instruments and in terms of common exposure to asset classes what leads to the emergence of systemic risk and makes the prediction of the behaviour of financial systems so difficult @xcite .
while risk diversification at a single institution can indeed lower its individual risk , if all institutions behave in a similar way , herding behaviour can instead amplify the risk . clearly , if all banks take similar positions , the failure of one bank can cause a global distress @xcite , because of the increased sensitivity to price changes @xcite . to add more complexity , the causes of market movements are still under debate @xcite , suggesting that exogenous instabilities add up to endogenous ones @xcite . the tension between individual regulation and global regulation @xcite poses a series of challenging questions to researchers , practitioners and regulators @xcite
traditionally , well before the recent crisis , it was argued that systemic risk is real when contagion phenomena across countries take place @xcite . in this spirit , a series of studies dealt with the description of systemic risk in the financial system from the perspective of the contagion channels across balance - sheet of several institutions @xcite . in particular ,
some focus was drawn upon the topology of connections ( or the network @xcite ) between institutions @xcite . in this way , the problem of analysing systemic risk splits in two distinct problems @xcite .
first , the problem of understanding the role of an opaque ( if not unknown ) structure of financial contracts @xcite and , second , the problem of providing a measure for the assessment of the impact of a given shock @xcite . as for the first problem , the obvious starting point is to consider the structure of the interbank network @xcite , with the aim of possibly extracting some early warning signals @xcite .
while many argued that the network structure can be intrinsically a source of instability , it turns out instead that no specific topology can be considered as systematically safer than the others @xcite .
indeed , only the interplay between market liquidity , capital requirements and network structure can help in the understanding of the systemic risk @xcite .
for the second problem , researchers have tried to describe the dynamics of propagation of defaults with various methods , including by means of agent - based models @xcite or by modelling the evolution of financial distress across balance - sheets conditional upon shocks in one or more institutions @xcite . from the perspective of financial regulations ,
capital requirements represent the cornerstone of prudential regulations .
institutions are required to hold capital as a buffer to shocks of any nature . the most used risk measures ( such as value at risk and expected shortfall )
are indeed related to the quantity of cash each individual bank needs to set aside in order to cover the _ direct _ exposures to different types of risk . in such manner ,
the _ indirect _ exposures arising from the interconnected nature of the financial system are not considered .
interconnectedness , though , is now entering the debate on regulation : for example , the definition of `` global systemically important banks '' ( g - sibs , * ? ? ?
* ) does include the concept of interconnectedness , thereby measured as the aggregate value of assets and liabilities each bank has with respect to other banking institutions .
although this represents a fundamental step towards the inclusion of interconnectedness in assessing systemic risk , a further level of disaggregation would be needed .
in fact , institutions that are similar in terms of their aggregated exposures ( including those _ vis - - vis _ other financial institutions ) , might have completely different sets of counterparties , therefore implying different levels of systemic impact and/or vulnerability to shocks .
another important point is that the potential negative effects arising from interconnectedness ought to be included into the definition of capital requirements .
in this section , we introduce and describe the debtrank stress - test framework .
one of the main characteristics of the framework lies in its _ flexibility _ along the following four main dimensions .
* shock type*. the framework can implement different shock types and scenarios ( on external assets ) .
* network estimation*. when detailed bilateral interbank exposures are not available , the framework provides a module to estimate the interbank network from the total interbank assets and liabilities of each bank , 3 .
* contagion dynamics*. the framework can implement two different contagion dynamics , distress contagion and default contagion .
* systemic risk indicators*. the framework returns as output a series of systemic risk indicators , both at the individual and a the global level .
the user can aptly combine this information to extract the information needed .
several graphical outputs are also available and represent a key feature of the framework : graphics are specifically designed to capture relevant information at a glance . given the flexibility of the framework and the number of outputs produced , in the remainder of the section , we focus on : 1 .
describing the main features of the debtrank _ distress process _ as the key foundation of the framework ; 2 . providing a qualitative description of the main variables of interests ; 3 . providing a technical summary of the _ building blocks _ of the framework , which include the _ inputs _ required , the _ outputs _ that can be obtained and the different _ modules _ constituting the framework
. the reader can find detailed information about the process and the main variables of interest in the methodological appendix [ sec : methods ] .
one of the key concerns in the measurement of systemic risk is to quantify _ losses _ at the individual and global level . in particular , debtrank focuses on the depletion of equity when banks experience losses in external or interbank assets .
we envision a system of @xmath2 banks ( indexed by @xmath3 ) and @xmath4 external assets ( indexed by @xmath5 ) .
the framework features a dynamic distress model , with @xmath6 : lll time & round & effects on balance sheets + @xmath7 & baseline & initial allocation + + @xmath8 & first round effects & .the distress dynamics.[tab : dynamics ] [ cols= " < " , ] +
the fundamental input data are represented by banks balance sheets . in particular
, the framework takes the equity , the total asset value and the total interbank lending and borrowing of each bank as minimal inputs .
more granular data on the structure of external assets are indeed possible ( e.g. in case one wants to simulate a shock on a specific asset class ) .
the flexibility of the modeling framework allows for a number of shock scenarios , including : 1 . a fixed shock ( e.g. @xmath9 ) on the value of all external assets ; 2 . a shock on the value of all external assets drawn from a specific probability distribution ( e.g. a beta distribution , which we use in the exercise in section [ sec : exercise ] . ) ;
when more detailed information on the holdings in external assets for banks is available , the shock ( either fixed or drawn from a probability distribution ) on specific asset classes .
as outlined above , the framework allows to compute the main systemic risk variable for two main types of contagion dynamics : 1 . the _ default cascade _ dynamics : banks impact other banks only in case of their default ( see , for the technical details , the discussion related to equation [ eq : main_debt_rank ] in the methodological appendix [ sec : methods ] . ) 2 . the _ debtrank _ dynamics : banks impact other banks regardless of whether the event of default occurred .
the rationale behind this type of dynamics is that , as banks reduce their equity levels to face losses , they decrease their distance to default and therefore are less likely to repay their obligations . in this case , the market value of their obligations is reduced and is hence reflected on the asset side of their counterparties in the interbank market .
as detailed data on banks bilateral exposures are often not publicly available , estimations need to be performed in order to run the framework .
even though such estimations constitute a key input of the stress test framework in case the exposures are not known , they constitute an output on their own , because they can be then analyzed with the typical tools of network analysis .
also , the estimations can serve for two other purposes : i ) as a benchmark for comparison with the observed data , _ la _
@xcite , or ii ) for the estimation of missing data @xcite . from a technical viewpoint , the methodology we use to estimate the interbank network
is based on the so called `` fitness model '' @xcite .
the technical details are reported in appendix [ sec : network_reconstruction ] .
in order to show how the framework works and what type of outputs are available , in this section we apply the framework to a specific dataset of @xmath10 eu banks for the years @xmath11 .
more details on the dataset are available in appendix [ sec : data_collection ] . in brief : 1 .
we collect yearly data on equity , external assets , interbank assets and liabilities for the set of banks under scrutiny ; 2 .
we estimated the exposures by combining the fitness model and an interative fitting procedure ( appendix [ sec : network_reconstruction ] ) , generating ( for each year ) @xmath12 networks compatible with the total interbank borrowing and lending of each bank at end - year ; 3 .
we then ran the stress - test in order to obtain the main systemic risk variables for all years .
when not explicitly specified , the statistics reported in this section are computed by taking the median value of the @xmath12 networks . in the remainder of this section , we describe the main results , including some key charts and figures , in order to show part of the graphical output of the framework .
figure [ fig : vulner_impact ] provides an overview of the response of the reconstructed financial networks and its individual elements to the distress scenarios simulated .
the chart on the left shows the dynamics of global equity losses ( @xmath13 ) from @xmath0 to @xmath1 , the values reported are the median value of @xmath13 across the @xmath12 networks in the monte carlo sample and are computed for a common shock of @xmath9 on the external assets .
the chart also offers a deconstruction of the losses , according to if they are caused by the first ( external assets shocks ) , second ( reverberation on the interbank lending network ) , and third ( fire sales ) round of distress propagation .
the relative losses in equity due to the second and third rounds are substantial , implying that an assessment of systemic risk solely based on first order effects is bound to underestimate potential losses .
the chart on the right shows the evolution of the impact for each of the @xmath10 banks in the sample throughout the years . each line is the median of the impact calculated over the @xmath12 networks in the ensemble .
the plot clearly shows a general decrease in the systemic impact for the individual institutions over time . in order to visually capture the persistency over time of banks with higher or lower impact ,
the colours reflect the level of the average impact computed over the years .
in particular , red lines are associated to banks that consistently show a high impact .
conversely , blue lines are associated to banks that have a consistently low impact .
we observe a certain level of stability of the relative levels : banks which show a higher systemic impact tend to do so throughout the years . from a systemic risk perspective
, it is of particular interest to compare the two main systemic risk quantities associated to each individual bank : the vulnerability to external shocks and the impact of a bank onto the system in case of its default . by _
jointly _ analyzing these two quantities , we divide institutions into four main categories : i ) high vulnerability / high impact , ii ) high vulnerability / low impact , iii ) low vulnerability / low impact , iv ) low vulnerability / high impact .
results for this exercise are reported in figure [ fig : vuln_vs_imp ] .
the graphs report a plot of the vulnerability @xmath14 at the second round versus the impact @xmath15 for each year in the sample .
the @xmath16 \times [ 0 , 1]$ ] square is divided into four quadrants , which correspond to the aforementioned four categories .
interbank leverage and total asset size are respectively visualised by node colour ( red implies high leverage , blue otherwise ) and node size . both interbank leverage and asset
size appear to be associated with high values of vulnerability and impact .
we observe an interesting phenomenon : in @xmath0 , a high number of large ( in terms of asset size ) institutions are _ both _ highly vulnerable ( up to their default ) and impactful ( up to @xmath17 of the total initial equity ) . their systemic relevance is therefore extremely high , as they have higher likelihood to receive distress . in turn , once the distress has been received , they would have a great impact on the rest of the system .
the situation improves over time and , in @xmath1 , no bank is in the upper right quadrant .
some financial institutions retain , though , very high vulnerability and significant impact .
a financial institution that can cause a global relative equity loss of 10% still acts as a source of systemic risk not to be ignored .
however , some large institutions are still prone to receive high level of distress , and nevertheless keep a significant impact ( up to @xmath18 on the rest of the system ) .
we also notice that those institutions which are both vulnerable and impactful are generally large and very large ones in terms of asset size .
figure [ fig : eff_decomp ] shows a way of visualizing the decomposition of first and second round effects .
again , we compare the years 2008 ( left ) and 2013 ( right ) .
the x - axis plots the losses at the first round and y - axis the losses after the second round . since the losses at the second round _ include _ the ones at the first , points must lie above the line bisecting the first quadrant .
nodes lying on the line itself are isolated in all the artificially generated networks .
we observe a significant reduction in the effects . as usual , the color reflects the interbank leverage and circle diameter the asset size . consistently with the findings in appendix [
sec : methods ] , nodes with higher interbank leverage typically suffer more losses in the second round .
we evaluate a distribution of relative global equity losses by simulating @xmath21 different systemic shock levels drawn from a beta distribution . and
@xmath22 respectively .
the distribution has been then truncated in order to attain a maximum value of @xmath23 and a minimum of @xmath24 . ]
figure [ fig : glob_losses_distr ] shows the distributions resulting by taking into account first only ( blue lines ) and second round ( red lines ) distress propagation effects for the years @xmath0 and @xmath1 .
vertical lines indicate var values at 95% , dashed lines are cvar at the same level ( see [ ssec : loss_distr ] for details ) .
an extremely important consideration can be made from this figure : accounting for second order effects greatly increases the likelihood of having larger global equity losses , thus _ shifting _ var values towards the right . in @xmath0 ,
a scenario where only first order distress is induced leads to a relatively low var level .
this , instead , reaches a much higher value after the second round effect is added . a similar ,
though less extreme , pattern is found in @xmath1 . the observed _ var shift _
phenomenon is another compelling piece of evidence stating that systemic risk measures ought to take into account network effects .
figure [ fig : individual_losses ] shows yet one of the outputs of the framework : the distribution of losses can be obtained for each individual bank .
here , we focus on two large institutions ( by asset size ) : hsbc ( which ranks first by asset size in @xmath1 ) and intesa sanpaolo ( which ranks thirteenth in @xmath1 ) . despite the difference in asset size , the original distance in the levels of var for the first round ( @xmath25 vs @xmath26 ) become much more relevant when second round effects are considered ( @xmath27 vs @xmath28 ) .
the example shows that significant differences in terms of standard risk measures are missed out if we neglect second - round effects .
the exercise carried out in section [ sec : exercise ] shows how the framework can be used to compute a variety of individual and global quantities that are relevant to systemic risk .
the framework allows for a number of additional analyses which are not reported in detail in this paper for the sake of conciseness .
for instance , figure [ fig : network_visualization ] represents one of the outputs of the framework in terms of network _ visualization _ and allows to compare the network position of individual institutions with other information . in this example , the interbank exposures among the top @xmath29 banks by total asset size in @xmath0 ( left ) and @xmath1 ( right ) are considered .
the position of a bank in the chart is determined by its impact : the higher the impact , the more central the bank is located in the circle .
the bubble size is proportional to total asset size of the bank , while the color encodes its vulnerability ( on a scale from blue to red , red nodes are more vulnerable ) .
it is worth mentioning the discussion on the determinants of the _ systemic importance of financial institutions_. in particular , one question is to what extent the asset size of an institution can be a good predictor of the impact of the bank on the system as a whole , and how much we should instead consider the position of the bank in the interbank network .
previous work have found that , although systemically important banks are typically among the large banks , banks with similar size can have very different impact on the system , in case of default @xcite . in line with those results , in our exercise , we find , loosely speaking , that asset size is _ not _ a good predictor of impact ( i.e. the pearson correlation between asset size and individual impact , as measured by debtrank , for the top @xmath30 institutions by total assets , each year is quite low , around @xmath31 ) . to summarize , this paper presents a stress - test framework focused on the evaluation of network effects in systemic risk .
we have illustrated how to carry out a stress - test exercise on a dataset of 183 european banks over the years @xmath0-@xmath1 .
the code underlying the framework has been developed in matlab and is available upon request to the authors .
the notion of interconnectedness has already entered the debate on `` global systemically important banks '' ( g - sibs , * ? ? ? * ) . however , this notion has been meant so far in an aggregate sense , without fully recognizing that institutions with similar aggregated exposures can have very different levels of systemic impact and/or vulnerability to shocks . indeed , a central notion in our framework is the one of _ leverage networks _ , i.e. the set of leverage relations among banks balance - sheets and among banks and assets .
the effect of these relations is the key starting point to monitor systemic risk from a network perspective .
accordingly , our framework allows to track separately the magnitude of the so - called first , second and third round effects , a feature that is particularly important in the discussion of future stress - tests at national and international level . in this respect , in line with previous work on german interbank data @xcite , we find that the second - round effect is at least as large as the first - round effect . notice that by adopting traditional analyses based on default - only mechanisms @xcite , we would instead find very limited second round effects .
there are two main reasons for this discrepancy .
first , in a default - only framework the propagation of distress only occurs in the case of outright default .
for example , in order to trigger any second round effect , at least one default in the first round is necessary : this would in turn imply that an initial shock needs to be at least as large as the reciprocal of a bank s leverage ( see appendix [ sec : methods ] ) . in order for the second round to be significant
, there would need to be a large number of defaults in the first round .
indeed , banks are recommended to keep their single largest exposure well below their capital so that a necessary condition for second round losses is the default of at least two of their counterparties .
however , in the practice , banks regularly re - evaluate their mark - to - market exposures to their counterparties in order to take into account the changes in their probability of default ( i.e. credit valuation adjustment ; see * ? ? ?
indeed , debtrank captures the effects of this adjustment in a recursive way .
the second reason for the discrepancy arises from the fact that , in @xcite , while the recovery rate on interbank assets is determined endogenously , the recovery rate on the external assets is assumed to be one .
although this assumption has been relaxed in @xcite , very large shocks at the first round are still necessary to trigger any second round . in contrast , the dynamics of debtrank assumes zero recovery rate on interbank assets , which is a realistic assumption in the short run .
future work will aim at bridging these two paradigmatic approaches . in the framework
, we further compute a series of systemic risk variables , along with their evolution over time , thus showing the dynamics of systemic risk in the financial system . in this respect , there is an added value in looking at quantities such as impact and vulnerability of financial institutions in combination , since systemic risk emerges when institutions that are systemically important become also vulnerable . while the results illustrated here have been obtained assuming the distress propagation mechanism of debtrank @xcite , other mechanisms can also be used in the framework and compared .
one of the obstacles in estimating network effects is the limitation in the availability of interbank exposures data . in order to address this issue ,
our framework allows to generate sets of interbank networks that satisfy the constraints on the total lending and borrowing of each bank . in this way
, we can gain insights on the possible range of variation on systemic risk , due to differences in key network quantities .
for example , one could tune the density parameter to assess whether this has an impact on the levels of systemic risk ( see appendix [ sec : network_reconstruction ] ) , or use various interbank network formation models ( see , e.g. * ? ? ? * ) .
overall , our aim is to enrich the set of existing tools by integrating the estimation of network effects with risk measures that are familiar to regulators and practitioners .
the most used risk measures ( such as value at risk and expected shortfall ) look at the buffer that each individual bank needs to set aside in order to cover the _ direct _ exposures to different types of shocks .
in contrast , the _ indirect _ exposures arising from the interconnected nature of the financial system are typically not considered in such measures . in this respect ,
our framework allows to estimate individual and aggregate banks loss distributions conditional to both direct shocks and indirect shocks on other banks .
in this methodological appendix , we provide the technical details of the process underlying the stress - test framework . in order to bridge between capital requirements and the network structure ,
we build on the common notion of _ leverage _ and define two _ leverage networks _ , which reflect a more granular representation of banks balance sheets . in the framework
, we consider a financial system composed of @xmath2 institutions ( banks ) .
each institution @xmath32 in the system can invest in either @xmath4 external assets or in the funding of the other @xmath33 financial institutions .
the focus of our analysis is on the dynamics of the balance sheets of each institution ( at each time @xmath34 ) and , in particular , of their equity levels .
the balance sheet is modelled as follows : @xmath35 is the equity value of institution @xmath32 at time @xmath36 , @xmath37 is value of its total assets and @xmath38 its total liabilities . consistently with much of the literature , we assume that assets are marked - to - market whereas liabilities are written at their face value
. we can classify assets and liabilities into _ external _ and _ interbank_. in particular , we consider the @xmath39 interbank lending matrix , whose element @xmath40 is the amount bank @xmath32 lends to bank @xmath41 in the interbank market and the @xmath42 external assets matrix , whose element @xmath43 is the amount invested by bank @xmath32 in the external asset @xmath44 .
the sum @xmath45 is the total amount of interbank assets of bank @xmath32 and the sum @xmath46 is the total amount of external assets of bank @xmath32 . in this framework , we consider external liabilities as exogenous and do not specifically model them : to simplify the notation , these liabilities do not carry a time index .
the balance sheet identity at each time @xmath47 reads : @xmath48 , or , equivalently , @xmath49 .
we define the _ total leverage _ of bank @xmath32 at time @xmath36 as the ratio between its total assets and its equity : @xmath50 , which can disaggregated into its additive subcomponents : @xmath51 where the element @xmath52 is the leverage of bank @xmath32 towards bank @xmath41 at time @xmath36 and the element @xmath53 is the external leverage of bank @xmath32 with respect to the external asset @xmath44 . by considering these two matrices as weighted adjacency matrices , we can then envision two _ leverage networks _ :
i ) a mono - partite interbank leverage network and ii ) a bipartite external leverage network . by summing along the columns of these matrices , we can obtain the _ total interbank leverage _
@xmath54 ( the interbank leverage _ out - strength _ ) and the _ total external leverage _
@xmath55 ( the external leverage _ out - strength _ ) .
these quantities are the key variables in our framework .
in particular , we will show that interbank and external leverage produce compounded effects when the dynamic of losses for the second round is considered . as banks deplete capital in order to face losses in both interbank and external assets , in the stress - test framework
we are mainly concerned with the dynamics of the relative loss in equity for each institution , with respect to a baseline level at @xmath47 .
this dynamics is captured by the following process : @xmath56 which represents the individual cumulative relative equity loss in time .
we assume that either no replenishment of capital or positive cash flow are possible , therefore @xmath57 , @xmath58 . in this way , the relative equity loss is a non - decreasing function of time .
further , @xmath59 $ ] @xmath58 . a bank defaults ( i.e. the bank reaches the maximum distress possible ) if @xmath60 . when @xmath61 the bank is undistressed .
all values of @xmath62 between @xmath63 and @xmath64 imply that the bank is under distress .
similarly , we can compute the global cumulative relative equity loss at each time @xmath36 as the weighted average of each individual level of distress : @xmath65 where the weights are given by @xmath66 , i.e. the fraction of equity of each bank at the baseline level ( @xmath47 ) . notice that @xmath62 is a pure number and so is @xmath67 .
the monetary value ( e.g. in euros or dollars ) of the loss can be obtained by @xmath68 ( individual loss ) and @xmath69 ( global loss ) . using the terminology introduced in the main text , equations [ eq : individual_equity_loss ] and [ eq : global_equity_loss ]
allow to measure the _ individual _ and _ global _ vulnerability respectively .
the entire distress process featured in the framework can be outlined in the following steps .
let @xmath70 be the value of one unit of the external asset @xmath44 . at time @xmath71 , a ( negative ) shock @xmath72 on the value of asset @xmath44 reduces the value of the investment in external assets of bank @xmath32 by the amount : @xmath73 .
banks record a loss on their asset side that , provided the hypothesis that assets are mark - to - market and liabilities are at face value , the loss needs to be compensated by a corresponding reduction in equity : @xmath74 the individual and global relative equity loss at time @xmath71 can be obtained as follows : ] ] @xmath75 which shows how the initial shock on each asset @xmath44 is _ multiplicatively _ amplified by the external leverage on that specific asset .
this leads to a straightforward interpretation of the leverage ratio .
indeed it is immediate to prove that the reciprocal of the leverage ratio corresponds to the minimum shock @xmath76 that leads bank @xmath32 to default ( this applies to all summands @xmath77 e @xmath78 in equation [ eq : summands ] ) .
since the single largest exposure is typically smaller than the equity , it is likely that defaults and large losses originate by different combinations of shocks affecting the different external assets . in the absence of detailed data on the exposure to different classes of external assets , we assume a _ common _ negative shock @xmath79 on the value of all external assets
. this assumption can be interpreted in two alternative ways .
first , we can envision a common small shock to all asset classes , as in times of general market distress .
the second way is that of a large shock to specific asset classes held by all banks ( e.g. sovereign on a class of countries , housing shocks , etc . ) . we can therefore drop the index @xmath44 in the summation and write : @xmath80 . at this point ,
the initial loss reverberates throughout the interbank network .
the debtrank algorithm @xcite extends the dynamics of default contagion into a more general distress propagation not necessarily entailing a default event .
in other words , shocks on the asset side of the balance sheet of bank @xmath32 transmit along the network even when such shocks are not large enough to trigger the default of @xmath32 .
this is motivated by the fact that , as @xmath32 s equity decreases , so does its _ distance to default _ @xcite and , consistently with the approach of @xcite the bank will be less likely to repay its obligations in case of further distress , therefore implying that the market value of @xmath32 s obligations will decrease as well .
consequently , the distress propagates onto its counterparties along the network .
if we denote the market value of the obligation with @xmath81 , is the element standing on the liability side of @xmath41 ( i.e. the face value established at time @xmath63 ) , whereas @xmath82 is the value ( mark - to - market ) at time @xmath36 written on the asset side of i. ] then above argument implies that the distress @xmath41 propagates onto its lender @xmath32 can be expressed , in general terms , as the relative loss with respect to the original face value @xmath83 . by summing over all obligors ,
the relative equity loss of each bank @xmath32 at time @xmath84 is described by : @xmath85 where @xmath86 is the set of _ active _ nodes , i.e. nodes that transmit distress at time @xmath36 .
the choice of the set of active nodes at time @xmath36 , @xmath86 , is a peculiarity of debtrank .
in fact , equation [ eq : main_debt_rank ] is of a recursive nature and therefore needs to be computed at each time @xmath36 by considering the nodes that were in distress at the previous time .
since the leverage network can present cycles , the distress may propagate via a particular link more than once .
although this fact does not represent a problem in mathematical terms , its economic interpretation is indeed more problematic . in order to overcome this problem
, debtrank excludes more than one reverberation . from a network perspective , by choosing the set @xmath86 we exclude walks that count a specific link more than once .
the process ends at a certain time @xmath87 , when nodes are no longer active .
[ [ the - functional - form - of - fcdot . ] ] the functional form of @xmath88 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the choice of the function @xmath88 deserves further discussion .
in fact , a correct estimation of its form would require an empirical framework which should take into account the probability of default of @xmath41 and the recovery rate of the assets held by @xmath32 .
however , the minimum requirement that @xmath88 needs to satisfy is that of being a non - decreasing relation between @xmath14 and the losses in the value of its obligations .
more specifically , we can hypothesize that small values of @xmath14 may have little to no effect on the market value of @xmath32 s obligations , whereas extremely large losses would settle the value of @xmath32 s obligations almost close to zero : the relationship is therefore necessarily non - linear and @xmath88 is likely to be a sigmoid - type of function .
in view of this , although further work will deal with the analysis of more refined functional forms , we hereby present two main forms , referring to the following two specific dynamics of distress : default contagion .
: : in this case , in line with a specific stream of literature , @xcite , only the event of default triggers a contagion .
the function @xmath88 is therefore chosen as the indicator function over the case of default @xmath89 .
: : the characteristics of @xmath88 imply the existence of an intermediate level where @xmath88 can be approximated by a linear function . by choosing the identity function @xmath90
, we obtain to the original debtrank formulation @xcite .
this functional form will be the one we use the most in the framework and the exercise . for the sake of clarity , in the remainder of this section ,
we consider only the latter functional form .
however , in the framework , stress tests can be easily carried out for both cases .
[ [ vulnerability . ] ] vulnerability .
+ + + + + + + + + + + + + + we are now ready to compute the vulnerability ( both individual and global ) and the _ impact _ ( at the individual level ) .
the individual vulnerability @xmath62 can be easily computed by setting @xmath91 in equation [ eq : main_debt_rank ] .
the global vulnerability is then given by @xmath92 . even though the framework can take as input _ any _ type of shocks ,
we focus briefly on the case in which the external assets of all banks are shocked : in this case all banks transmit distress at time @xmath71 and , given the choice of the set @xmath93 , the process indeed ends at time @xmath94 .
we can hence derive a closed - form solution for the individual vulnerability after the second round : @xmath95 which elucidates the _ compounding _ effect of external and interbank leverage .
if the shock @xmath96 is small enough not to induce any default , then [ eq : first_order_approx ] can be rewritten as : @xmath97 [ [ impact.-1 ] ] impact .
+ + + + + + + debtrank , in its original formulation @xcite , entails a stress test by assuming the default of each bank individually and computing the global relative equity loss _ induced _ by such default .
this is indeed what we define as the _ impact _ of an institution onto the system as a whole .
formally , this can be written as : @xmath98 [ [ network - effects - a - first - order - approximation - of - vulnerability ] ] network effects : a first order approximation of vulnerability + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + equation [ eq : main_debt_rank ] clearly shows the main feature of the distress dynamics captured by debtrank : the interplay between the network of leverage and the distress imported from neighbors in this network .
further , equation [ eq : first_order_approx ] clarifies the multiplicative role of leverage in determining the distress at the end of the second round .
we now develop a first - order approximation of equation [ eq : first_order_approx ] , which will serve the purpose of further clarifying the _ compounding _ effects of external and interbank leverage in determining distress .
for the sake of simplicity , we assume no default , which allows us to remove the `` @xmath99 '' operator .
this is a reasonable assumption in case of a relatively small shock on external assets .
we approximate the external leverage of the obligors of bank @xmath32 by taking the weighted average ( with weights @xmath100 ) of their external leverages , which we denote by @xmath101 . as @xmath102
, we write @xmath103 . by denoting with @xmath104 the weighted average of @xmath105
, we can approximate the global equity loss at the end of the second round @xmath106 as : @xmath107 which allows to see how the second - round effects alone can be obtained as the product of the weighted average interbank leverage and weighted average external leverage .
typically , stress tests emphasize the effects of the first - round : as we observe , this may potentially bring to a severe underestimation of systemic risk .
after the second round , banks have experienced a certain level of equity loss that has completely reshaped the initial configuration of the balance sheets at time @xmath47 .
banks are now attempting to restore , at least partially , this initial configuration .
in particular , we assume @xcite that each bank @xmath32 will try to move to the original leverage level @xmath108 .
this implies that banks will try to sell external assets in order to obtain enough cash to repay their obligations and therefore reduce the size of their balance sheet .
because of the vast quantity of external assets sold by the banking system in aggregate , the impact on the prices of external assets is also relevant , which will reduce accordingly .
banks therefore will experience further losses due to fire sales and we label such losses as _ third round _ effects . here , we provide a minimal model for the scenario described above .
consider the leverage dynamics at @xmath109 .
the leverage at @xmath36 is @xmath110 we assume that , at @xmath47 , each bank had a quantity of external assets @xmath111 and , without loss of generality , that the initial price of the asset is unitary ( @xmath112 ) . hence , the asset values at @xmath47 can be written as @xmath113 .
the asset price after the first round is therefore simply @xmath114 .
recalling that the first round affects only the external asset and that the second round affects only interbank assets , the leverage of each bank @xmath32 immediately after the second round can be written as : @xmath115 where , for ease of notation @xmath116 and @xmath117 .
first , we need to prove that the new leverage levels are higher with respect to the initial conditions .
it is easy to prove that @xmath118 ( as long as @xmath32 has not defaulted ) : @xmath119 where @xmath120 .
the above inequality leads to the condition @xmath121 , which is always verified in our setting . at @xmath122 , banks attempt to restore the target leverage @xmath123 , by selling a fraction @xmath124 $ ] of their external assets at the price @xmath125 and replenish their equity of an amount @xmath126 .
therefore , we modify equation [ eq : newleveragelevels ] as follows : @xmath127 after some passages , we obtain the value for @xmath128 : @xmath129 which satisfies equation [ eq : newnewleverage ] .
the relative amount of assets sold is given by @xmath130 .
we further assume that the simultaneous selling of external assets in the market produces a further linear impact on the price .
given the impact of fire sales , the new price is further reduced as follows : @xmath131 and the relative change in price is therefore proportional to the relative change in quantity of sold assets through a constant @xmath132 $ ] .
finally , by computing the additional loss given by the decline in price following equation [ eq : pricedecline ] , we obtain the final individual relative equity loss at @xmath133 : @xmath134 and the global equity loss at the third round ( assuming no defaults ) : @xmath135 the distress process allows to capture , at each time @xmath36 , the relative equity loss for both the individual institution and the system as a whole .
this implies the possibility to compute , at each time @xmath36 , a ( continuous ) _ relative equity loss distribution _
conditional to a certain shock .
the equity loss distribution can be characterized , for example , by two typical risk measures : value at risk ( var ) and conditional value at risk ( cvar ) ( also known as expected shortfall , es ) . since @xmath62 and @xmath67 are nonnegative variables @xmath136 $ ] @xmath137 , the individual value at risk for bank @xmath32 at time @xmath36 at level @xmath138 is defined as the @xmath139 quantile @xcite : @xmath140 : p(h_i(t ) \leq x ) \geq ( 1 - \alpha)\}\ ] ] and the conditional value at risk for bank @xmath32 at time @xmath36 at level @xmath138 is defined as the expected value of the losses exceeding the var , as : @xmath141\ ] ] considering the system as a whole , we can likewise analyze the global relative equity losses @xmath67 at each time @xmath36 , therefore obtaining a _ global var _ : @xmath142 : p(h(t ) \leq x ) \geq ( 1 - \alpha)\},\ ] ] and the _ global cvar _ : @xmath143.\ ] ]
detailed public data on banks balance sheets are unavailable , therefore we resorted to a dataset that provides a reasonable level of breakdown , the bureau van dijk bankscope database ( url : ` bankscope.bvdinfo.com ` ) .
we focus on a subset of @xmath10 banks headquartered in the european union that are also quoted on a stock market for the years from 2008 to 2013 .
the main criterion for the selection was that of having detailed coverage ( on a yearly basis ) for total assets , equity , interbank lending or borrowing .
future work will deal with data at higher frequency ( quarterly , monthly , ) .
our interbank asset and liability data include amounts due under repurchase agreements ( which are economically analogous to a secured loan ) thereby prompting large contagion effects .
we perfomed a series of consistency checks . in the case of missing interbank lending data for a bank for less than three years
, we proceed with an estimation via linear interpolation of the data available for the other years ( a comparison with the available data gives errors lower than 20% ) .
since , in general , the correlation between interbank lending and borrowing for all banks and years is about 70% ( with some significant differences ) , this implies the presence of net lenders and net borrowers . in view of this , when data on either interbank lending or borrowing are not available for more than three years , we simply set them equal .
data on total interbank lending and borrowing are often publicly available , while the detailed bilateral exposures are typically confidential . however , in this section , we outline the estimation procedure adopted in the framework . at each point in time , we create a sample of @xmath12 networks via the `` fitness model '' , which is a technique that has recently been used to reconstruct financial networks starting from aggregate exposures @xcite . the procedure can be outlined as follows : * 1 .
total exposure re - balancing .
* since we are considering a subset of the entire interbank market , we observe an inconsistency : the total interbank assets @xmath144 are systematically smaller than the total interbank liabilities @xmath145 for each year ( eu banks are net borrowers from the rest of the world ) . to adopt a conservative scenario , we assume that the total lending volume in the network is the minimum between the two ( @xmath146 in the exercise ) .
let @xmath147 and @xmath148 be respectively the lending and borrowing propensity of @xmath32 .
* 2 . exposure link assignment .
* the fitness model , when applied to interbank networks @xcite attributes to each bank a so - called fitness level @xmath149 ( typically a proxy of its size in the interbank network ) .
we can estimate the probability that an exposure between @xmath32 and @xmath41 exists via the following formula , @xmath150 ( @xmath151 is a free parameter ) .
notice that @xmath152 .
consistently with a recent stream of literature @xcite , for each bank we take as fitness @xmath149 the average between its total lending and borrowing propensity , implying that , the greater this value , the higher will be the number of counterparties ( the _ degree _ of a node ) . considering empirical evidence on the density of different interbank networks @xcite , we assume on average a density of @xmath153 ( i.e. about @xmath154 over the @xmath155 possible links ) . does not influence the overall results of the exercise .
for example , values for the global vulnerability at the second round differs only at the third decimal digit . ]
since it can be proved that the total number of links is equal to the expected value of @xmath156 , we can determine the parameter @xmath151 and compute the matrix of link probabilities @xmath157 .
we now generate @xmath12 network realizations . for each of these realizations ,
we assign a link to the pair of banks @xmath158 with probability @xmath157 .
the link direction ( which determines whether @xmath32 or @xmath41 is the lender or the borrower ) is chosen at random with probability @xmath31 . *
exposure volume allocation * last , we need to assign weights to the edges ( the volumes of each exposure ) .
we impose the fundamental constraint that the sum of the exposures of each bank ( out - strength ) equals its total interbank asset @xmath159 . to achieve this
, we implement an iterative proportional fitting algorithm on the interbank exposure matrix @xmath160 .
we wish to estimate the matrix @xmath161 , which is the relative value of each exposure with respect to the total interbank volume .
we begin the estimation @xmath162 of @xmath163 , at each iteration : ( 1 ) @xmath164 , i.e. @xmath162 is divided by its relative lending propensity and multiplied by the total relative assets of @xmath32 , ; ( 2 ) @xmath165 @xmath166 .
we repeated the two steps until @xmath167 and @xmath168 are below @xmath9 .
last , the exposure network can be estimated by @xmath169 . | we develop a novel stress - test framework to monitor systemic risk in financial systems .
the modular structure of the framework allows to accommodate for a variety of shock scenarios , methods to estimate interbank exposures and mechanisms of distress propagation .
the main features are as follows .
first , the framework allows to estimate and disentangle not only _ first - round effects _ ( i.e. shock on external assets ) and _ second - round effects _
( i.e. distress induced in the interbank network ) , but also _ third - round effects _ induced by possible fire sales .
second , it allows to monitor at the same time the _ impact _ of shocks on individual or groups of financial institutions as well as their _ vulnerability _ to shocks on counterparties or certain asset classes .
third , it includes estimates for loss distributions , thus combining network effects with familiar risk measures such as var and cvar .
fourth , in order to perform robustness analyses and cope with incomplete data , the framework features a module for the generation of sets of networks of interbank exposures that are coherent with the total lending and borrowing of each bank . as an illustration , we carry out a stress test exercise on a dataset of listed european banks over the years @xmath0-@xmath1 .
we find that second - round and third - round effects dominate first - round effects , therefore suggesting that most current stress - test frameworks might lead to a severe underestimation of systemic risk .
stefano battiston,^a^ guido caldarelli,^b^ marco derrico,^a,^ stefano gurciullo^c^ + ^a^ department of banking and finance , university of zurich + ^b^ imt alti studi lucca , isc - cnr , rome , lims london + ^c^ school of public policy , university college london + this version : |
quantum antiferromagnets have provided one of the most fertile grounds for discoveries of new states of matter and testing the theoretical models . the one - dimensional heisenberg antiferromagnetic chain of @xmath6 spins ( 1d - haf ) is probably the most studied example of such quantum antiferromagnets where recent advances in the theoretical approaches@xcite have enabled quantitative discussion of experimental data on model systems . among the best realizations of 1d - hafs are kcuf@xmath9,@xcite sr@xmath1cuo@xmath9,@xcite cuso@xmath10d@xmath1o,@xcite cu(c@xmath11h@xmath11n@xmath1)(no@xmath9)@xmath1,@xcite copper - pyrimidine - dinitrate@xcite and cuse@xmath1o@xmath12.@xcite excellent agreement between the extensive experimental and theoretical work established the existence of gapless @xmath13 excitations , the so called spinons,@xcite into which the conventional @xmath14 magnon excitations fractionalize .
the low - energy description of gapless phases in 1d - hafs leads to the tomonaga - luttinger liquid ( tll ) paradigm@xcite as a universal concept of quantum physics in one dimension , beyond the perturbation theory .
in all the above - cited realizations of 1d - hafs , the spin states stem from the electrons in @xmath15orbitals of transition metal ions .
recently , an important structural and magnetic susceptibility investigation of cso@xmath1 , a member of the alkali superoxide _
a_o@xmath1 ( _ a _ = alkali metal ) family,@xcite suggested this compound to realize the first @xmath8orbital 1d - haf.@xcite 1d - haf state is established in the orbitally - ordered phase below @xmath2 k on cooling , where a spatial alignment of the @xmath0 molecular orbitals of o@xmath16 units with @xmath6 results in a strong superexchange interaction @xmath17 along the crystallographic @xmath18axis , i.e. , the chain axis , while in the perpendicular crystallographic @xmath19direction the exchange interaction is very small due to poor overlap of the corresponding molecular orbitals ( fig .
[ fig_struct ] ) .
comprehensive @xmath20cs nuclear magnetic resonance ( nmr ) measurements of spin - lattice relaxation rate @xmath21 in different magnetic fields indeed revealed the characteristic power law dependence @xmath22,@xcite which is a hallmark of the tll state.@xcite the field - dependence of the extracted parameters quantitatively demonstrates that the low - energy spectrum of spin - fluctuations in the 1d - haf state of cso@xmath1 can be described in terms of tll yielding a dimensionless tll exponent @xmath23 , consistent with the ising - like exchange - coupling anisotropy . because the proposed 1d - haf state in cso@xmath1 is based on @xmath8orbital states that occupy a different regime of interplay between magnetism , structure , vibrations , and orbital ordering than the more conventional @xmath24orbital or @xmath25orbital states , finding additional independent experimental evidences for the 1d - haf state is highly desirable .
crystal structure and relevant orbitals .
zig - zag chains of ordered o@xmath26 @xmath27 molecular orbitals ( @xmath6 ) run along the crystallographic @xmath18axis .
blue and red spheres represent caesium and oxygen atoms , respectively .
the superexchange interactions , @xmath17 , between @xmath27 orbitals ( yellow ) are bridged by cs @xmath28 orbitals ( green ) . ]
electron paramagnetic resonance ( epr ) is a resonance technique that has been broadly used to study low - dimensional quantum antiferromagnets.@xcite traditionally , temperature dependence of the shift and linewidth of the epr spectra have been discussed with the spectral moment analysis , i.e. , the so - called kubo - tomita theory.@xcite however , this approach was suggested to be inappropriate in the case of 1d - hafs .
instead , oshikawa and affleck employed the sine - gordon quantum - field theory to circumvent some computational difficulties of the quantum spin state in 1d - hafs and to predict the temperature dependence of the epr linewidth for different magnetic anisotropies.@xcite experimental tests on model systems showed an excellent agreement with this theory.@xcite interestingly , exact low - energy solutions for epr spectra in terms of tll exist in the spin channel only for metallic wires@xcite and are still missing for the case of insulating spin - chain systems .
first epr studies of cso@xmath1 single crystals were reported already back in the 1970 s by knzig,@xcite but clearly the experimental data need to be reinterpreted in light of the new findings .
therefore , we decided to remeasure the x - band epr spectra and to test the data against the oshikawa - affleck and the tll theories .
an excellent agreement is found with both theoretical models below @xmath3 , thus providing an independent support for the 1d - haf state in cso@xmath1 .
therefore , cso@xmath1 indeed represents a rare example of the @xmath8orbital one - dimensional quantum antiferromagnet .
the cso@xmath1 powder sample came from the same batch as the one used in our previous nmr study where the preparation was described in detail.@xcite freshly destilled cs metal was oxidized with @xmath29o@xmath1 gas . to homogenize the yellow product of cs@xmath29o@xmath1
, several cycles of grinding under a controlled argon atmosphere and temperature annealing in natural o@xmath1 was processed .
the sample was enriched by the @xmath29o isotope to approximately 50% , according to raman spectroscopy .
laboratory x - ray diffraction at room temperature confirmed high purity of the sample , as all peaks could be indexed to the tetragonal , @xmath30 , structure of cso@xmath1 .
no impurity phase could be traced down to the resolution of our xrd equipment .
the magnetic properties of the sample , which were first checked by squid magnetometry,@xcite are in good agreement with previous results.@xcite in particular , an inflection in the magnetic susceptibility curve @xmath31 occurring near 70 k on heating ascribed to the structural / orbital ordering transition and a broad hump near 30 k implying the 1d - haf state , are clearly visible . a curie - weiss analysis of the high - temperature @xmath32 data obtained in a field of 0.5 t yields an effective magnetic moment @xmath33 of @xmath34 ( here ,
@xmath35 is bohr magneton ) and a curie - weiss constant @xmath36 k. similarly as in ref .
, @xmath33 is considerably larger than the spin - only value of @xmath37 for an @xmath13 system , thus confirming an orbital contribution to @xmath33 in the high temperature phase of cso@xmath1 . the nel ordering transition below 10 k
is mostly obscured by a low - temperature curie tail due to a small amount of paramagnetic impurities .
however , a weak feature in @xmath38 around 9 k in data measured at low magnetic field of 0.01 t indicates the transition to the ordered state . for the epr measurements ,
23.1 mg of powder sample was sealed under dynamic vacuum in a standard suprasil quartz tube .
the x - band epr experiments were performed on a home - built spectrometer equipped with a varian e-101 microwave bridge , a varian tem104 dual cavity resonator , an oxford instruments esr900 cryostat and an oxford instruments itc503 temperature controller .
the temperature stability was better than @xmath39 k at all temperatures .
the epr spectra were measured on cooling .
_ temperature evolution of the x - band epr spectra . _
the cso@xmath1 powder sample shows no detectable x - band epr signal at room temperature . on cooling ,
a very broad line starts to appear below @xmath40 k and becomes very pronounced below @xmath3 ( fig .
[ slika_spektri_cezij ] ) . for temperatures close to @xmath3 ,
the epr spectrum is shifted towards higher fields relative to @xmath41 , thus indicating an orbital angular momentum contribution in the high - temperature orbitally - disordered phase and consistent with the large high - temperature @xmath33 extracted from the squid data.@xcite however , as temperature decreases well below @xmath3 , the epr signal narrows and shifts to @xmath41 resonance - field range .
the powder spectra are well fitted with the uniaxial @xmath42factor anisotropy , i.e. , @xmath43 , convoluted with a lorentzian lineshape broadening whose linewidth also has the uniaxial anisotropy , i.e. , @xmath44 . at @xmath45 k
, an unconstrained fit converges to @xmath46 and @xmath47 , which agree with the literature data obtained on single crystals.@xcite moreover , since we consistently find @xmath48 at all temperatures , we next constrain @xmath49 to find that @xmath50 nearly linearly decreases with decreasing temperature between 60 and @xmath51 k. the epr spectrum is the narrowest at @xmath52 k where @xmath53 mt . on further cooling , the signal broadens again as the nel temperature @xmath54 k is approached from above before it completely disappears below @xmath5 and is replaced by another slightly asymmetric signal shifted to significantly lower fields .
for example , at @xmath55 k this new resonance is centered at around 80 mt . since this signal
is found only below @xmath5 , we tentatively assign it to one of the antiferromagnetic resonance modes . at lowest temperatures we notice another extremely weak peak at @xmath56 , which we attribute either to impurity o@xmath16 centers or to point defects in the antiferromagnetically ordered magnetic lattice .
powder ( grey lines ) . dashed black lines are fits with the uniaxial @xmath57-factor and linewidth anisotropy ( @xmath43 and @xmath58 ) .
solid red lines are fits with eq .
( [ tll_spekter ] ) to the tll spectrum . below the nel temperature
@xmath4 k the paramagnetic epr signal completely disappears and antiferromagnetic resonance ( marked by * ) is observed at around 80 mt at @xmath59 k. at @xmath60 k we observe also an extremely weak impurity line at @xmath56 ( marked by o ) . ] _ linewidth analysis with the oshikawa - affleck theory .
_ as recent structural and nmr investigations of cso@xmath1 interpreted the phase that develops below @xmath3 as a 1d - haf phase,@xcite we proceed to discuss the measured epr spectra ( fig .
[ slika_spektri_cezij ] ) in terms of the one - dimensional magnetism .
we start with the oshikawa - affleck theoretical framework of a low - temperature epr response in @xmath6 one - dimensional heisenberg antiferromagnets , which predicts distinct temperature dependences of epr linewidth for the case of anisotropic exchange broadening , where @xmath61 , as opposed to the case of staggered - field broadening , where @xmath62.@xcite since we find that @xmath50 nearly linearly increases with increasing temperature between 15 k and @xmath3 ( fig .
[ slika_sirina]a ) , we conclude that the main anisotropic interaction that broadens epr spectra is the symmetric anisotropic exchange interaction .
a rapid increase in @xmath50 on cooling between 15 k and @xmath63 is attributed to the critical fluctuations just before the three - dimensional antiferromagnetic order sets in . in the temperature range between @xmath64 and @xmath63
, the temperature dependence of @xmath50 is thus fitted to @xmath65 which comprises linewidth contributions from the 1d - haf phase with symmetric anisotropic exchange@xcite and from the critical fluctuations , respectively . here , @xmath66 is the boltzman constant
. the constant @xmath67 amounts to 2 for magnetic field directed along the anisotropy axis and to 1 for the perpendicular case , thus nicely accounting for the experimental observation @xmath68 . since we have defined @xmath69 , we use @xmath70 in eq .
( [ oa_fit ] ) .
an excellent fit of the experimental epr linewidth ( fig .
[ slika_sirina]a ) yields the constant of symmetric anisotropic exchange @xmath71 .
although we can not determine the sign of @xmath72 , because the corresponding linewidth term is proportional to @xmath73 , we note a significant magnetic anisotropy that is consistent with the nmr finding of the ising - like exchange - coupling anisotropy .
the magnitude of the critical fluctuations is found to be @xmath74 t and the critical exponent @xmath75 .
_ analysis of the epr spectra within the tomonaga - luttinger - liquid framework . _
the above analysis demonstrates that the epr data of cso@xmath1 may indeed be interpreted with the 1d - haf model , at least for @xmath76 .
however , a recent @xmath20cs nmr study of cso@xmath1 employed the tll formalism instead,@xcite so that it is difficult to directly compare the two results .
therefore , for the quantitative comparison with the nmr data we next test our epr data also against the tll model .
the epr spectrum @xmath77 is given by the imaginary part of the dynamical spin susceptibility for the transverse direction ( i.e. , perpendicular to the magnetic field ) , @xmath78 , calculated at the momentum of the excitation @xmath79 and at the resonant frequency @xmath80 , namely @xmath81.@xcite here , @xmath82 is the magnitude of the perpendicular ( excitation ) microwave field , @xmath83 the permeability of the vacuum and @xmath84 the volume of the sample .
we stress at this point that the epr spectrum provides complementary information to the nmr spin - lattice relaxation rate , which is proportional to @xmath78 integrated over the brillouin zone . at low frequencies ,
the dynamical spin susceptibility has nonzero contributions in the vicinity of the four @xmath85 vectors in the brillouin zone , i.e. , @xmath86 and @xmath87 for the transverse direction , and @xmath88 and @xmath89 for the longitudinal direction ( i.e. , along the magnetic field ) . here
@xmath90 is the field - induced magnetization , which is related to the two tll parameters , i.e. , the dimensionless exponent @xmath91 and the velocity of spin excitations @xmath92 , via @xmath93 .
the first and the third of the listed modes are commensurate , whereas the other two modes are field - dependent ( through @xmath90 ) and thus incommensurate . among these
, the epr spectrum is determined by the transverse mode with nonzero contribution close to @xmath79 , i.e. , by the mode at @xmath94 .
for this mode the dynamical spin susceptibility takes the form @xcite @xmath95\ , , \end{aligned}\ ] ] where @xmath96 and @xmath97 is small . here , @xmath98 is the non - universal amplitude,@xcite @xmath99 is the euler s beta function and @xmath100 . to satisfy the epr resonance condition
@xmath79 , we have to set @xmath101 , which finally leads to @xmath102.\end{aligned}\ ] ] the signal - intensity prefactor @xmath103 is proportional to the electronic static spin susceptibility @xmath104 , to @xmath82 and @xmath84 . the arguments @xmath105 and the function @xmath106 $ ] .
we stress that eq .
( [ tll_spekter ] ) is analogous to the one calculated for the spin channel of the correlated one - dimensional metals , e.g. , carbon nanotubes.@xcite fitting all the experimental x - band epr spectra ( fig .
[ slika_spektri_cezij ] ) in the temperature range @xmath107 to @xmath108 with @xmath103 and @xmath91 as the only free parameters produces excellent lineshape fits ( solid red lines in fig .
[ slika_spektri_cezij ] ) yielding @xmath109 .
we note that large broadening of the epr spectra is found as soon as @xmath91 deviates significantly from @xmath110 , which is again reminiscent of the case of interacting itinerant electrons in one dimension.@xcite the temperature dependence of the epr signal intensity is shown in fig .
[ slika_sirina]b .
the intensity shows a broad maximum at @xmath111 k and is rapidly suppressed upon approaching the nel temperature from above .
we note that @xmath112 matches the temperatures of the maxima in the bulk susceptibility measurements@xcite and in the local spin susceptibility probed by @xmath20cs nmr.@xcite the best fit of the epr signal intensity to the 1d - haf spin susceptibility@xcite in the temperature range from 60 k to 15 k is found for @xmath113 k , which is close to the corresponding @xmath114 k extracted from the magnetic susceptibility@xcite and @xmath20cs nmr data.@xcite the deviations of the fit from the data points close to @xmath64 are expected due to a large lineshape broadening and lineshift in this temperature range ( fig .
[ slika_spektri_cezij ] ) , implying underestimated data values , whereas the discrepancies below @xmath115 k are due to the onset of three - dimensional antiferromagnetic correlations .
the temperature dependence of the epr signal intensity and the magnitude of the extracted @xmath17 thus fully corroborate the proposed tll state below @xmath64 .
spectra ( open black circles ) .
solid red line represents fits to eqs .
( [ oa_fit ] ) or ( [ tll_fit ] ) , which are undistinguishable . dashed blue line shows the contribution of the 1d - haf phase to the total epr linewidth .
( b ) temperature dependence of the x - band epr signal intensity of cso@xmath1 ( open black circles ) .
solid red line is fit to the static spin susceptibility of the 1d - haf@xcite with @xmath113 k. dotted vertical black lines mark the nel ( @xmath116 k ) and orbital - ordering temperatures ( @xmath117 k ) . ] finally , since in our case @xmath118 and @xmath119 , we can by using arguments of ref .
simplify eq .
( [ tll_spekter ] ) to a sum of two lorentzian components , one centered at the positive and the other one at the negative resonance field , with the temperature - dependent linewidth @xmath120 not surprisingly , in this limit the temperature dependence of the epr linewidth takes , in terms of the temperature dependence , the same form as eq .
( [ oa_fit ] ) . therefore ,
fitting the experimental linewidth to @xmath121 and using the same parameters for critical fluctuations @xmath122 and @xmath123 as above , we find @xmath109 consistent with the value of @xmath91 obtained from the epr lineshape fits . to solve the problem of antiferromagnetically coupled @xmath124 spins in one dimension , both the oshikawa - affleck and the tll theories bosonize the spin operators and then use field - theory approaches .
however , to calculate the epr response , the older oshikawa - affleck theory@xcite calculates the epr width and shift perturbatively .
the exact solutions for the dynamic spin susceptibilities were published only later@xcite and here we take a full advantage of these newly made discoveries to calculate exactly the epr spectrum . given the same starting point for the oshikawa - affleck and the tll approaches , it is thus not surprising that both theories equally well describe the epr spectra in the parameter range relevant for cso@xmath1 .
the most important finding of this work is thus that the temperature dependence of the epr lineshape , the epr linewidth , and the epr signal intensity fully comply with the interpretation of the orbitally - ordered phase of cso@xmath1 in terms of tll framework , with @xmath91 consistently converging to the value close to but yet slightly smaller than @xmath110 .
the extracted @xmath91 and the symmetric anisotropy - exchange @xmath72 places cso@xmath1 on the ising - anisotropy side of the phase diagram .
although this conclusion in general agrees with the previous analysis of the @xmath20cs nmr data@xcite , we note that @xmath125 derived from the present epr data is notably larger than @xmath23 extracted from the high - field ( @xmath126 t ) nmr experiments on the same sample .
this discrepancy remains puzzling .
however , since epr and nmr probe the low - energy part of @xmath127 in a slightly different way , the features in the excitation spectrum arising from the effects of interchain couplings may echo differently in epr and nmr data .
the origin of the symmetric anisotropic exchange is unclear at the moment , but most likely it directly reflects the fact that the cs@xmath128 ions , characterized by a significant spin - orbit coupling constant , are bridging the superexchange between nearest neighboring o@xmath26 units along the @xmath18axis .
in conclusion , a systematic analysis of the x - band epr spectra in the low - temperature orbitally - ordered phase of cso@xmath1 agrees with the proposed 1d - haf state . using either the oshikawa - affleck or the tll theoretical frameworks , we find a systematic agreement with the experimental epr lineshape , linewidth and signal intensity , thus allowing us a quantitative determination of the anisotropy of the symmetric exchange interaction or , alternatively , of the tll exponent . therefore
, our results indeed comply with cso@xmath1 as a @xmath8orbital system showing a quantum antiferromagnetic state where orbital ordering is pivotal in the formation of @xmath6 spin chains . since similar orbital - ordering physics is ubiquitous to the whole family of _
superoxide compounds@xcite and _ a_@xmath11o@xmath129@xcite sesquioxide compounds , it would be interesting to test in the future the low - dimensional quantum magnetism also in these systems .
d.a . and c.f .
acknowledge the financial support by the european union fp7-nmp-2011-eu - japan project lemsuper under contract no .
we thank walter schnelle and ralf koban for performing magnetization measurements .
y. kono , t. sakakibara , c. p. aoyama , c. hotta , m. m. turnbull , c. p. landee , and y. takano , _ phys .
lett . _ * 114 * , 037202 ( 2015 ) .
s. a. zvyagin , a. k. kolezhuk , j. krzystek , and r. feyerherm , _ phys .
lett . _ * 95 * , 017207 ( 2005 ) . | recently it was proposed that the orbital ordering of @xmath0 molecular orbitals in the superoxide cso@xmath1 compound leads to the formation of spin-1/2 chains below the structural phase transition occuring at @xmath2 k on cooling . here
we report a detailed x - band electron paramagnetic resonance ( epr ) study of this phase in cso@xmath1 powder .
the epr signal appears as a broad line below @xmath3 , which is replaced by the antiferromagnetic resonance below the nel temperature @xmath4 k. the temperature dependence of the epr linewidth between @xmath3 and @xmath5 agrees with the predictions for the one - dimensional heisenberg antiferromagnetic chain of @xmath6 spins in the presence of symmetric anisotropic exchange interaction .
complementary analysis of the epr lineshape , linewidth and the signal intensity within the tomonaga - luttinger liquid ( tll ) framework allows for a determination of the tll exponent @xmath7 .
present epr data thus fully comply with the quantum antiferromagnetic state of spin-1/2 chains in the orbitally ordered phase of cso@xmath1 , which is , therefore , a unique @xmath8orbital system where such a state could be studied . |
our understanding of the early universe has improved considerably in recent years through the photometric discovery of large numbers of high - redshift galaxies in both deep and gravitationally - lensed fields observed with the _ hubble space telescope _ ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although uncertainties remain , the demographics and limited spectroscopic follow - up of this early population has been used to argue that star - forming galaxies played a significant role in completing the reionization of the igm by redshift @xmath11 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a key observation delineating the end of the reionization epoch is the marked decline beyond @xmath126 in the visibility of lyman - alpha ( ly@xmath0 ) emission seen in continuum - selected galaxies @xcite . as ly@xmath0 represents a resonant transition it is readily scattered by the presence of neutral gas and thus acts as a valuable proxy for the state of the igm .
[ fig1 ] there are , however , several limitations in using ly@xmath0 as a probe of reionization .
firstly , converting the declining visibility of ly@xmath0 into a neutral gas fraction involves complex radiative transfer calculations and several uncertain assumptions . the currently observed decline ( e.g. * ? ? ?
* ) implies a surprisingly rapid end to the process @xcite .
secondly , simulations suggest that reionization is likely to be a patchy process @xcite and thus conclusions drawn from the modest samples of spectroscopically - targeted galaxies may be misleading @xcite .
finally , the absence of a ly@xmath0 detection in the spectrum of a proposed high - redshift candidate may simply imply the source is a foreground galaxy ( although the contamination rate for such galaxies is typically very low , e.g. @xcite ) .
ideally targets in such studies would be spectroscopically confirmed independently of ly@xmath0 , for example using uv metal lines ( * ? ? ?
* ; * ? ? ?
* see also @xcite ) .
although observational progress is challenged by the faintness of targets selected in deep fields such as the hubble ultra deep field ( typically @xmath1327 - 28 ) , an important development has been the identification of much brighter @xmath14 candidates from the wider area , somewhat shallower , cosmic assembly near - infrared deep extragalactic legacy survey ( candels , * ? ? ?
* ; * ? ? ?
surprisingly , some of these brighter targets reveal ly@xmath0 despite lying inside the putative partially neutral era .
* hereafter ) reported ly@xmath0 with a rest - frame equivalent width ( ew ) of 8 at @xmath15=7.508 in a @xmath16=25.6 galaxy ; ( * ? ? ?
* hereafter ) find ly@xmath0 emission at @xmath15=7.73 with ew=21 in an even brighter source at @xmath16=25.03 ; and ( * ? ? ?
* hereafter ) identified a tentative ly@xmath0 emission ( 4.7@xmath17 ) in a @xmath16=25.12 galaxy at a redshift @xmath15=7.477 , which we have now confirmed ( stark et al , in prep ) .
in addition to their extreme luminosities ( @xmath18 ) , these three sources have red [ 3.6]-[4.5 ] spitzer / irac colors , indicative of contamination from strong [ o iii ] and balmer h@xmath19 emission . using the multi - object spectrometer for infra - red exploration ( mosfire , @xcite ) on the keck 1 telescope , we report the detection of a prominent emission line in a further bright candidate drawn from the candels program .
egsy-2008532660 ( hereafter egsy8p7 ; ra=14:20:08.50 , dec=+52:53:26.60 ) is a @xmath16=25.26 galaxy with a photometric redshift of @xmath20 and a red irac [ 3.6]-[4.5 ] color , recently discovered by .
we discuss the likelihood that the line is ly@xmath0 at a redshift @xmath21 making this the most distant spectroscopically - confirmed galaxy .
detectable ly@xmath0 emission at a redshift well beyond @xmath22 raises several questions regarding both the validity of earlier claims for non - detections of ly@xmath0 in fainter sources , and the physical nature of the luminous sources now being verified spectroscopically .
even if these bright systems are not representative of the fainter population that dominate the ionization budget , they offer new opportunities to make spectroscopic progress in understanding early galaxy formation .
the paper is organized as follows : in [ obs ] we review the object selection , spectroscopic observations , and data reduction .
the significance of the line detection and its interpretation as ly@xmath0 is discussed in [ lya ] .
we discuss the implications of the detectability of ly@xmath0 in the context of the earlier work in [ discussion ] .
throughout we use a standard @xmath23cdm cosmology with @xmath24 , @xmath25 , @xmath26 @xmath27 km s@xmath8mpc@xmath8 , @xmath28 , and magnitudes are given using the ab convention .
errors are @xmath29 unless otherwise stated . [ fig2 ]
the galaxy egsy8p7 was detected in the extended groth strip ( egs ; @xcite ) from deep ( @xmath30 ) multi - band images in the candels survey and first reported as one of four unusually bright ( @xmath31 ) candidate @xmath14 galaxies by .
one of these , egs - zs8 - 1 , with @xmath32 was spectroscopically confirmed at @xmath33 by .
normally such objects would be selected as @xmath34-band dropouts but , given @xmath34-band observations are not yet available over the full candels field , an alternative selection criterion was adopted that takes advantage of red irac [ 3.6]-[4.5 ] colors , indicating prominent [ o iii]+h@xmath19 emission within the 4.5 @xmath35 m band ( see also * ? ? ?
* ; * ? ? ?
of the objects listed by , egsy8p7 is the only one for which y - band data is not available . as a result ,
its photometric redshift derived from _
data alone is fairly uncertain ( @xmath36 ; fig . 5 of ) but
including its irac [ 3.6]-[4.5 ] color of [email protected] ( and blue @xmath38-[3.6 ] color ) narrows the range to @xmath1 .
we observed egsy8p7 , the highest - redshift candidate in s list , on june 10 and 11 2015 with mosfire on the keck 1 telescope .
observations in the j - band spanned the wavelength range @xmath39 using an ab dithering pattern of @xmath40 along the slit with individual frames of 120s .
the slit masks on the two nights differed by 120@xmath41 in orientation both with a slit width of @xmath42 . in each mask
we allocated one slit to a nearby star to monitor changes in seeing , transparency , and possible positional drifts .
conditions were clear throughout , with an average seeing of @xmath43 for the first night and @xmath44 for the second night .
the total nightly exposure times were 158 minutes and 128 minutes . excluding exposures where the seeing was significantly ( @xmath45 ) worse than average
, the useful exposure time comprises all frames from the first night and 80% of those acquired on the second night ( where the conditions were less stable ) culminating in a total exposure - time of 4.33 hours .
calibrations were obtained via long - slit observations of standard a0v stars .
data reduction was performed using the standard mosfire reduction pipeline . for each flat - fielded slit we extracted the 1d spectrum using a 11-pixel boxcar centered on the expected position of the target .
a similar procedure was adopted in quadrature to derive the 1@xmath17 error distribution . to ensure that the derived error spectrum reflects the noise properties of the data , we measured the standard deviation of the pixel - by - pixel snr , which should be unity for a set of independent values drawn from gaussian distributions .
we obtained a value of 1.3 , implying that the errors are slightly underestimated by the pipeline , and corrected the error spectrum accordingly .
data from both nights were co - added by inverse - variance averaging the calibrated 1d spectra .
we also allowed for relative shifts along the slit of 2 pixels ( @xmath40.2 ) and across the slit of 0.2(which affects the expected slit loss correction ) and propagated the associated uncertainty into our error budget .
all reductions and calibration steps were performed independently by two authors ( az , sb ) . to calibrate the spectra we scaled a vega model to each standard star to determine a wavelength - dependent _ relative _ flux calibration and telluric correction using the procedure described in @xcite .
independent telluric calibrations derived from three standard stars agree to within @xmath46 at the location of the detected emission line .
the _ absolute _
calibration for each night was derived by comparing the spectroscopic magnitude of the stars on the slitmask with photometric measures in the @xmath47-band obtained from the 3d - hst catalog @xcite .
note that this procedure takes into account differential slit losses due to varying seeing .
the absolute calibration factors obtained for the two nights are consistent with differences arising from seeing effects .
we adopt a conservative absolute calibration error of @xmath48 , taking into account the nightly variation , differences between the independent reductions , and other contributions mentioned above . the final 2d and 1d spectra
are shown in fig .
fig . [ fig1 ] reveals a prominent line at @xmath49 flanked by a skyline on its blue side .
examining the signal and associated noise in a circular aperture of 6 pixels in radius corresponding to @xmath50 times the line width found below
we estimate a significance of @xmath51 in the 2d spectrum and , within the extracted 1d spectrum over the same spectral range , @xmath52 .
this significance holds over a range of integration wavelengths both including and excluding the skyline .
additionally , the line is detected independently on each night ( @xmath53 on night 1 , @xmath54 on night 2 ) , despite the changed slit orientation ( fig .
[ fig2 ] ) . taking the signal / noise on the first night , and assuming it scales as @xmath55/fwhm , where @xmath56 is the exposure and fwhm the seeing , for the second night ( where the exposure was less and seeing worse ) we predict a signal / noise similar to that observed .
we estimate the observed line properties by using a monte carlo markov chain to fit a truncated gaussian to the data ( in a similar way to ) , taking into account instrumental broadening .
we leave the truncation point a free parameter , and mask out pixels contaminated by strong skyline residuals .
the proximate skyline significantly affects the fit , and different masking configurations yield different results . to span the range of possible solutions we repeat the model fitting , as well as direct integration measurements , with a variety of masking configurations . in each such configuration
a different part of the skyline was excluded . for each of these resulting models
we calculate the 95% confidence level intervals . for each parameter
we then take the union of the intervals from different models and report them in table 1
. the line may be asymmetric with some attenuation on its blue side , but this is difficult to ascertain given the proximate skyline , and the best - fit models ( fig .
[ fig1 ] ) are mostly symmetric .
the line peak lies between 11770 and 11776 which , for ly@xmath0 , implies a redshift of @xmath57 .
note the typical peak wavelength is a few bluer than the raw peak in fig .
the line width ranges from 2.08.4 , corresponding to a velocity broadening , corrected for instrumental effects , of v@xmath58=51214 km s@xmath8 .
the line flux ranges from 1.02.5@xmath7 erg s@xmath8 @xmath9 ( excluding the 30% uncertainties in the absolute calibration ) .
_ importantly , the significance of the line remains at the @xmath597@xmath17 level for all chosen integration methods and variations in the masking of the skyline_. the reliability of the snr measurement given the nearby skyline was further verified using simulated lines inserted into the 2d spectra .
lll name & fiducial model & 95% c.i .
+ + f(ly@xmath0 ) & 1.7 & [ 1.0 - 2.5 ] + @xmath35 & 11774 & [ 11770 - 11776 ] + z & 8.683 & [ 8.679 - 8.684 ] + ew & 28 & [ 17 - 42 ] + @xmath17 & 4.7 & [ 2.0 - 8.4 ] + @xmath60 & 118 km s@xmath8 & [ 51 -214 ] km s@xmath8 + @xmath61 & 277 km s@xmath8 & [ 120 -503 ] km s@xmath8 + [ tab : specsummary ] we can estimate the chance of finding such a line by examining the signal / noise distribution in our combined 2d spectral data at 1500 random locations taking the same aperture used for measuring the line properties above . the resulting histogram in fig .
[ random ] reveals a very low probability of finding such a feature by chance.our measured signal / noise represents a 6@xmath17 deviation in the distribution ( which has a width larger than unity due to correlated noise ) .
we also verified that proximity to skylines does not increase the width of the distribution . together with the facts that the positive and negative line positions in the 2d spectrum match the dithering pattern used ( fig.[fig1 ] ) and the line position is astrometrically centered on the target position along the slit on both nights , we conclude the probability for the detection to be an artifact is negligibly small . .
all apertures were 6 pixels in radius ( @xmath62 in wavelength space ) , as the aperture used to measure the line significance .
the vertical line shows the detected signal / noise far exceeds that of the noise distribution.,width=340 ] [ random ] adopting the identification of ly@xmath0 , the inferred redshift of @xmath63 is satisfactorily close to the photometric estimate , @xmath1 reported by . using the _ hst _ @xmath64-band photometry unaffected by line emission , for a reasonable uv slope ( @xmath65 )
, the range of line fluxes ( table 1 ) translates to a rest - frame ew of 1742 . while formally fairly uncertain , it is comparable to line properties secured with mosfire for the other bright galaxies at @xmath10 selected in candels with red irac colors ( , ) .
obtained a total line flux of @xmath66 erg s@xmath8 @xmath9 and a rest - frame ew of @xmath67 , whereas secured a total line flux of @xmath68 erg s@xmath8 @xmath9 and a more modest rest - frame [email protected] .
we note discussed the possibility that egsy8p7 is magnified by foreground galaxies . adopting the lens galaxy masses found in as input , and using an updated photometric redshift ,
we conclude egsy8p7 is likely magnified by @xmath69 , and less than a factor @xmath50 .
given we only detected one spectral line in the j - band , we also considered interpretations other than ly@xmath0 , corresponding to a lower redshift galaxy . if the line is a component of the [ o ii ] ( @xmath703726 , 3729 ) doublet , this would imply @xmath71 , possibly consistent with a low redshift solution to the sed ( fig .
[ fig4 ] ; @xmath72 ; 99% c.i . ) .
however , for each [ o ii ] component we find no trace of the other line to a limit several times fainter than the detected line ( _ magenta _ arrows in fig . [ fig1 ] ) .
similarly , in the case of [ o iii ] ( @xmath704959 , 5007 ; _ cyan _ arrows ) , or the h@xmath73 line ( @xmath74 ) , implying @xmath75 , we would expect to see ( at least ) one of the other lines given typical line ratios assumed .
while the presence of skylines clearly limits a robust rejection of these alternatives , both the absence of any optical detections and the red irac color are hard to understand in a low redshift galaxy with intense line emission .
formally , the sed fit , performed using the eazy program @xcite , gives a low redshift likelihood @xmath76 times smaller than our adopted solution at @xmath77 ( fig .
[ fig4 ] ) .
upper - limits from non - detections .
blue _ sed shows the best - fit template at @xmath15=8.57 .
the _ magenta _ sed shows the @xmath15=8.48 solution adopting the strong emission line template used in ( see their work for more details on the templates ) .
the _ grey _ sed shows the best - fit low redshift template which can not explain the absence of detection in the optical bands and which is @xmath78 times less likely than the high - redshift solutions.,width=340 ]
although we claim egsy8p7 is the highest - redshift spectroscopically - confirmed galaxy and the first star - forming galaxy secured beyond a redshift of 8 ( c.f . a gamma ray burst at @xmath79 , @xcite ) , its prominent ly@xmath0 emission , detected in only 4 hours , raises several interesting questions about the usefulness of galaxies as tracers of reionization .
the last few years has seen a consistent picture emerging from observational programs charting the visibility of ly@xmath0 beyond a redshift @xmath126.5 in much fainter ( @xmath80 ) galaxies ( @xcite and references therein ) . although the exposure times , selection techniques and observational strategies differ across the various programs , almost 100 faint targets have now been targeted , collectively yielding very few @xmath14 ly@xmath0 detections ( , @xcite ) . in marked contrast , all 3 bright @xmath10 targets in the egs field in table 2 of s compilation now reveal ly@xmath0 emission .
although this bright spectroscopic sample is admittedly modest , the high success rate in finding ly@xmath0 at redshifts where the igm is expected to be substantially neutral ( @xmath470% according to @xcite ) raises two interesting questions .
firstly , do these new results challenge claims made by the earlier fainter ly@xmath0 searches in the same redshift interval ?
is it possible that those campaigns reached to insufficient depths to reliably detect ly@xmath0 emission ? typically such studies adopted the rest - frame ew distribution at @xmath126 published by @xcite and simulated , using their appropriate survey and instrumental parameters , the likely success rate to somewhat higher ew limits of @xmath425 - 50 .
a key issue in understanding whether the objects are similarly attenuated by the partially neutral igm at @xmath81 is the fiducial @xmath126 ew distribution for these luminosities .
stark et al s ew distribution is based on an analysis of 74 @xmath82 galaxies targeted spectroscopically within the deep goods fields but only 13 reached the uv luminosities being probed here .
@xcite present a smaller sample of 10 spectroscopically - confirmed @xmath126 galaxies from the uds field , several of which are more luminous with ews @xmath59 40 .
a key deficiency in this comparison is that the @xmath82 samples were not selected to have as large ew [ oiii]+h@xmath19 emission implied in the irac [ 3.6]-[4.5 ] color - selected sample .
@xcite show that ly@xmath0 ews are uniformly extremely large ( 40 160 ) in @xmath122 - 3 galaxies with such large ew optical line emission .
thus it is possible that ly@xmath0 in these @xmath10 galaxies is attenuated .
in addition to the possibility of cosmic variance in the egs field , it may be that such luminous objects at @xmath14 lie in large overdense regions and present a somewhat accelerated view of the reionization process @xcite .
the second question is whether these rarer luminous sources ( c.f .
@xcite ) are physically distinct from the fainter population .
adopting the irac red [ 3.6]-[4.5 ] color selection may preferentially select unusually intense line emission with abnormal ionizing spectra capable of creating early bubbles of ionized gas in the local igm ( see also , @xcite ) .
evidence for stronger than usual ionizing spectra in @xmath14 dropouts with known ly@xmath0 emission is provided by the prominent detection of civ 1548 emission in the lensed galaxy at @xmath83 @xcite .
the key issue in this case is the proportion of @xmath84 lyman break galaxies that reveal such prominent irac excesses .
@xcite demonstrate that perhaps upwards of 50% of the photometrically selected galaxies at @xmath126.6 - 6.9 have irac excesses that are comparable in magnitude to those seen in . in summary , not only
have we pushed the detectability of ly@xmath0 emission to well beyond a redshift @xmath128 , but it seems luminous @xmath85 galaxies selected additionally for their red irac [ 3.6]-[4.5 ] colors have an unusually high success rate of line detection .
we present several reasons why this might contrast with the low yield of detecting ly@xmath0 in intrinsically fainter sources .
regardless of the explanation , it is clear that further spectroscopic follow - up of such examples will yield valuable information on the early development of massive galaxies .
we thank the reviewer of this work for useful comments . we acknowledge useful discussions with jim dunlop , brant robertson and chuck steidel .
we thank garth illingworth , pieter van dokkum , iva momcheva , marijn franx , mauro stefanon , and benne holwerda for their role in the selection of this object .
support for this work was provided by nasa through hubble fellowship grant # hst - hf2 - 51334.001-a awarded by stsci , which is operated by the association of universities for research in astronomy , inc . under nasa contract nas 5 - 26555 .
rse acknowledges the hospitality of the institute for astronomy , edinburgh via a carnegie centennial professorship .
this work is in part based on previous observations made with the nasa / esa hubble space telescope , the spitzer space telescope , the subaru telescope , and the canada france hawaii telescope .
the data presented herein were obtained at the w.m .
keck observatory .
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 . , i. s. , steidel , c. c. , epps , h. w. , et al .
2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 8446 , society of photo - optical instrumentation engineers ( spie ) conference series , 0 | we report the discovery of lyman - alpha emission ( ly@xmath0 ) in the bright galaxy egsy-2008532660 ( hereafter egsy8p7 ) using the mosfire spectrograph at the keck observatory .
first reported by @xcite , it was selected for spectroscopic observations because of its photometric redshift ( @xmath1 ) , apparent brightness ( h@xmath2 ) and red spitzer / irac [ 3.6]-[4.5 ] color indicative of contamination by strong oxygen emission in the [ 4.5 ] band . with a total integration of @xmath34.3 hours ,
our data reveal an emission line at @xmath411776 which we argue is likely ly@xmath0 at a redshift @xmath5 , in good agreement with the photometric estimate .
the line was detected independently on two nights using different slit orientations and its detection significance is @xmath6 .
an overlapping skyline contributes significantly to the uncertainty on the total line flux although the significance of the detected line is robust to a variety of skyline - masking procedures . by direct addition and a gaussian fit
, we estimate a 95% confidence range of 1.02.5@xmath7 erg s@xmath8 @xmath9 , corresponding to a rest - frame equivalent width of 1742 .
egsy8p7 is the most distant galaxy confirmed spectroscopically to date , and the third luminous source in the egs field beyond @xmath10 with detectable ly@xmath0 emission viewed at a time when the intergalactic medium is believed to be fairly neutral .
although the reionization process was probably patchy , we discuss whether luminous sources with prominent irac color excesses may harbor harder ionizing spectra than the dominant fainter population thereby creating earlier ionized bubbles .
further spectroscopic follow - up of such bright sources promises important insight into the early formation of galaxies . |
in @xcite , detemple showed that the isosceles triangle of smallest perimeter which circumscribes a semicircle is made up of two congruent right triangles which have sides proportional to @xmath2 , and the right triangle is known as the kepler triangle@xcite@xcite . here , we ll consider another interesting problem that determine a triangle of smallest area which circumscribes a semicircle . in figure
[ fig : fg1a ] , let a triangle @xmath3 circumscribe a semicircle of radius @xmath4 , with the diameter of the semicircle contained in the base @xmath5 , let @xmath6 denote the center of the semicircle , @xmath7 and @xmath8 are both the radii , then let @xmath9 and @xmath10 , now , our problem is : [ pm ] if the radius @xmath11 and the two sides @xmath12 and @xmath13 of @xmath3 can not be three edges of a triangle , determine @xmath3 of smallest area .
[ prop1 ] if @xmath11 , @xmath12 and @xmath13 can not be three edges of a triangle , then the of smallest area @xmath3 is a right triangle similar to the ( @xmath14 , @xmath15,@xmath16 ) triangle which forms half of a golden rectangle@xcite@xcite@xcite .
if @xmath11 , @xmath12 and @xmath13 can not be three edges of a triangle , then first , we suppose the radius @xmath4 is the longest segment , thus , we have @xmath17 , where @xmath18 , then in @xmath3 , we have the area equation ( [ 1 ] ) @xmath19 and @xmath20 @xmath21 then we get @xmath22 hence , we conclude that the radius @xmath11 can not be the longest segment among the three segments .
next , we suppose @xmath13 is the longest segment , then we have @xmath23 , using the similar method , we can get @xmath24 since @xmath25 , we have @xmath26 and with @xmath27 , we conclude that @xmath28 , where @xmath29 , hence @xmath30 the equality in ( [ 1d ] ) holds if and only if @xmath31 .
then we get @xmath32 , @xmath33 , and in ( [ 1b ] ) , we have @xmath34 , which means @xmath3 is a right triangle having sides proportional to ( @xmath14 , @xmath15,@xmath16 ) .
, scaledwidth=100.0% ] there is an identity ( [ 2 ] ) of the golden ratio@xcite and fibonacci numbers ( see , e.g. , @xcite ) .
@xmath35 if we rewrite ( [ 2 ] ) in the form of ( [ 2a ] ) , @xmath36 we will obtain a right triangles sequence @xmath0 with sides ( @xmath14 , @xmath37 , @xmath38 ) , see figure [ fig : fg2 ] .
it s easy to see that , the first right triangle @xmath39 with sides ( @xmath14 , @xmath40 , @xmath15 ) is the kepler triangle whose side lengths are in geometric progression , the second right triangle @xmath1 is a ( @xmath14 , @xmath41 , @xmath42 ) triangle , and furthermore , let @xmath43 , we find that the limiting right triangle of @xmath0 is just a ( @xmath14 , @xmath15 , @xmath16 ) triangle which forms half of a golden rectangle .
since there are only two golden ( see @xcite ) right triangles ( one is the kepler triangle with sides ( @xmath14 , @xmath40 , @xmath15 ) , the other is the ( @xmath14 , @xmath15 , @xmath16 ) triangle ) , and they are both special cases of @xmath0 , then , we can call @xmath0 a generalized golden right triangles sequence .
in addition , we have the following simple area inequality ( [ 2b ] ) for @xmath0 , @xmath44 where @xmath45 denoted as the area of @xmath0 , and @xmath46 in the inequality ( [ 2b ] ) , we also notice that the second triangle @xmath1 with sides ( @xmath14 , @xmath41 , @xmath42 ) is just the maximum area triangle in @xmath0 , therefore , we can call @xmath1 the maximum generalized golden right triangle .
interestingly enough , similar to the construction of the kepler triangle @xmath39 by first creating a golden rectangle shown in @xcite , we can also construct the maximum generalized golden right triangle @xmath1 by first constructing a golden rectangle , see figure [ fig : fg2a ] . 1 .
construct a golden rectangle @xmath47 with @xmath48 , @xmath49 ( see , e.g. , @xcite ) 2 .
construct @xmath6 dividing @xmath5 in the golden ratio and @xmath50 3 .
draw an arc with the center at @xmath6 and the radius of length @xmath13 , cutting the extension of @xmath51 at @xmath52 , and join @xmath52 to @xmath53 [ pmlst ] if the radius @xmath11 and the two sides @xmath12 and @xmath13 of @xmath3 can not be three edges of a triangle , determine @xmath3 of smallest perimeter . and if @xmath3 can not be an acute - angled triangle , determine @xmath3 of smallest perimeter .
d. w. detemple , the triangle of smallest perimeter which circumscribes a semicircle , _ the fibonacci quarterly _ , * 30.3 * ( 1992 ) 274 .
l. mario , _ the golden ratio : the story of phi , the world s most astonishing number _ , broadway books , new york , 2002 .
kepler triangle , _
wikipedia , the free encyclopedia _ , https://en.wikipedia.org/wiki/kepler_triangle .
t. koshy , _ fibonacci and lucas numbers with applications _ , a wiley - interscience publication , 2001 .
m. bicknell and v. e. hoggatt , golden triangles , rectangles , and cuboids , _ the fibonacci quarterly _ , * 7.1 * ( 1969 ) 7391 .
a. s. posamentier and i. lehmann , _ the fabulous fibonacci numbers _ ,
prometheus books , 2007 . | an interesting problem that determine a triangle of smallest area which circumscribes a semicircle is solved . then a generalized golden right triangles sequence @xmath0 is obtained , and an interesting construction of the maximum generalized golden right triangle @xmath1 is also shown . |
uniaxial ferromagnets composed of earth - abundant elements are of interest to researchers exploring new materials for permanent magnet applications @xcite .
fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 are examples of such materials that have not been heavily investigated in this respect to date . the crystal structures and compositions of these compounds were first investigated by aronsson and coworkers @xcite and rundqvist @xcite . they were shown to be isostructural , adopting the tetragonal cr@xmath0b@xmath3 structure type .
ferromagnetism was first reported in the phosphide based on magnetization measurements @xcite and mssbauer spectroscopy @xcite , which identified the uniaxial nature of the magnetic structure with the easy axis of magnetization along the crystallographic _
c_-axis . soon thereafter ,
fe@xmath0sib@xmath1 was identified as a uniaxial ferromagnet at room temperature as well @xcite , and a spin reorientation was seen to occur near 140 k , resulting in planar magnetic anisotropy at lower temperature @xcite .
the curie temperatures ( @xmath4 ) of both phases were found to be well above room temperature , near 630 k for fe@xmath0pb@xmath1 @xcite and 784 k for fe@xmath0sib@xmath1@xcite .
a very recent study of fe@xmath0pb@xmath1 reports @xmath4 = 655 k , magnetic anisotropy comparable to hard ferrites , and a magnetic moment per fe of 1.72 @xmath5 @xcite . the isostructural mn compounds are also ferromagnetic , but with curie temperatures closer to room temperature , making them perhaps more interesting as magnetocaloric rather than magnet materials @xcite .
no reports of magnetization measurements for fe@xmath0sib@xmath1 were located in the literature , and no detailed crystallographic study has been published to date . in addition ,
understanding of effects of chemical substitutions in these uniaxial , high - temperature ferromagnets is presently lacking . in this work , we report structural and magnetic properties determined from measurements on polycrystalline samples of fe@xmath0pb@xmath1 , fe@xmath0sib@xmath1 , fe@xmath2copb@xmath1 , fe@xmath2cosib@xmath1 , fe@xmath2mnpb@xmath1 , and fe@xmath2mnsib@xmath1 .
full crystal structure refinements for all compounds are reported at room temperature , and for both fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 at 20 k. significant anisotropy is observed in the thermal expansion , which is negative along the _ c_-direction in fe@xmath0sib@xmath1 .
magnetization measurements are used to determine the saturation magnetizations and identify the curie and spin - reorientation temperatures .
results of density functional theory calculation of the ordered magnetic moment and anisotropy of fe@xmath0sib@xmath1 and fe@xmath2mnsib@xmath1 are reported , and compared to the measurements .
the silicides are found to have a larger magnetic moment than the phosphides , but smaller magnetic anisotropy at 300 k is indicated in the silicides .
both chemical substitutions suppress the spin reorientation in fe@xmath0sib@xmath1 , indicating fe@xmath2cosib@xmath1 and fe@xmath2mnsib@xmath1 are uniaxial ferromagnets down to at least 30 k. in both fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 , substitution of mn expands the unit cell , has little effect on the curie temperature , and produces a slight reduction in magnetic moment .
substitution of co contracts the unit cell , and strongly suppresses both the curie temperature and the magnetic moment .
these findings suggest that the ordered magnetic moment on co in these structures is either very small or aligned non - collinearly with the net magnetization .
polycrystalline samples of nominal compositions listed in table [ tab : properties ] were synthesized by reacting the elements in evacuated and flame - sealed silica tubes .
the tubes were heated to 1000 @xmath6c at 100@xmath6c per hour , and held at this temperature for 12 hours .
the resulting mixtures were then ground into powders and pressed into 1/2 inch diameter pellets at room temperature .
the pellets were sealed in evacuated silica tubes and heated at 1000@xmath6c for one week .
the pellets were then ground into powder , re - pelletized , and heated in evacuated silica tubes at 1000@xmath6c for 5 days .
samples produced in this way were found to be @xmath790% pure .
for the silicon - containing samples ( with no phosphorus ) , the initial mixing of the elements can be done by arc - melting , but subsequent grinding , pelletizing , and annealing are needed to improve the phase purity .
this later route was used to make the fe@xmath2cosib@xmath1 used in this study , because it gave a final sample with less ferromagnetic impurity .
x - ray diffraction was performed on a panalytical x - pert pro diffractometer using monochromatic cu k@xmath8 radiation in bragg - brentano reflection geometry .
an oxford phenix cryostat was used for diffraction measurements below room temperature . quantitative analysis and crystal structure refinement
was done with rietveld refinement using the program fullprof @xcite . to determine the easy axis of magnetization by diffraction ,
finely ground samples were sprinkled onto a vacuum grease coated glass slide with a strong permanent magnet underneath .
compositions of the primary phases were examined using energy dispersive analysis with a bruker quantax 70 x - ray spectrometer ( eds ) in a hitachi tm3000 tabletop scanning electron microscope .
magnetization measurements on solid polycrystalline samples cut from the sintered pellets were performed with a quantum design magnetic property measurement system .
analysis of powder x - ray diffraction patterns ( fig .
[ fig : pxrd ] ) shows that all of the materials studied adopt the tetragonal cr@xmath0b@xmath3 structure type at room temperature .
the samples used in the current study were @xmath790% pure , and the impurities identified in the diffraction patterns along with their weight percent concentrations are shown on the figure . in the fe@xmath0pb@xmath1-based materials , the only secondary phase detected was fe@xmath1p , which is paramagnetic at room temperature , and undergoes a first - order ferromagnetic phase transition near 216 k @xcite .
about 3 % of fe@xmath9si@xmath1b @xcite was observed in the fe@xmath0sib@xmath1 and fe@xmath2mnsib@xmath1 samples .
the magnetic properties of this phase have not been reported , but in the course of this study we have synthesized and performed preliminary characterization of a sample of composition fe@xmath10si@xmath1b and find that it is likely paramagnetic at room temperature and antiferromagnetic at lower temperature . the fe@xmath2cosib@xmath1 sample ( fig .
[ fig : pxrd]e ) contains a small amount of an unidentified impurity phase , which may be reflected in the high temperature magnetization data presented below .
[ cols= " < , > , > , > , > , > , > " , ] powder x - ray diffraction patterns from fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 were also collected at 20 k. the results from both materials were well described by the room temperature structures , with lattice parameters 5.4766(1 ) and 10.3488(2 ) for the phosphide and 5.5412(1 ) and 10.3390(2 ) for the silicide .
other crystallographic parameters determined at 20 k are shown in table [ tab:20k ] .
a crystallographic distortion may be expected to occur at the spin reorientation temperature in fe@xmath0sib@xmath1 ; however , no distortion was resolved in the data , indicating the magnetoelastic coupling in the _
ab_-plane is not strong .
comparison of lattice constants at room temperature ( table [ tab : properties ] ) and 20 k ( table [ tab:20k ] ) reveal strongly anisotropic behavior . in fe@xmath0pb@xmath1 , the contraction of _ a _ upon cooling from room temperature to 20 k is more than four times larger than the relative contraction along _
c_. in fe@xmath0sib@xmath1 , the basal plane also changes more strongly with temperature than the @xmath11-axis .
in fact , the @xmath11-axis shows negative thermal expansion in this compound .
the thermal expansion was examined further by monitoring the 130 and 006 bragg reflections upon cooling from 300 to 20 k. the results are shown in figure [ fig : thermal - expansion]a .
the position of the 130 reflection depends only on the basal plane lattice constant _ a _ and the 006 position depends only on the _ c_-axis length .
normal behavior is seen in fe@xmath0pb@xmath1 .
the negative thermal expansion along _ c _ is apparent in the temperature dependence of the position of the 006 reflection in fe@xmath0sib@xmath1 .
this compound is known to undergo a spin reorientation below room temperature , with the easy axis reported to be in the _ ab_-plane at low temperature @xcite .
results of magnetization measurement with the field applied approximately along the easy axis of magnetization are shown in figure [ fig : aligned - pxrd]b .
for these measurements , finely ground powders were mixed in epoxy and cured at room temperature in a magnetic field .
the spin reorientation is apparent near 170 k in fe@xmath0sib@xmath1 .
as expected , no similar anomaly is seen in fe@xmath0pb@xmath1 , which remains uniaxial at low temperatures @xcite .
in addition , the mn and co substitutions studied here appear to suppress the spin reorientation in fe@xmath0sib@xmath1 .
no lattice anomalies are apparent in figure [ fig : aligned - pxrd]a at the spin reorientation temperature in fe@xmath0sib@xmath1 .
magnetization vs temperature measured in an applied field of 0.01 t for the p - containing compounds ( a ) and the si - containing compounds ( b ) .
the data were collected on cooling and are normalized to the values measured at 400 k. curie temperatures estimated from the data are noted on the figure .
, width=312 ] high temperature magnetization measurements were conducted to determine the curie temperature ( @xmath4 ) of each material .
the results are shown in figure [ fig : jvst ] .
the measurements were performed in a relatively low applied field of 0.01 t to obtain sharp changes in @xmath12 at @xmath4 .
the @xmath4 values are listed on figure [ fig : jvst ] and in table [ tab : properties ] , and were determined by extrapolating to @xmath13 = 0 the steepest part of the magnetization curve .
relatively small contributions to the magnetization arising from ferromagnetic impurity phases are seen in fe@xmath0pb@xmath1 ( fig .
[ fig : jvst]a ) , fe@xmath2cosib@xmath1 , and fe@xmath2mnsib@xmath1 ( fig .
[ fig : jvst]b ) .
the identities of the additional ferromagnetic phases is not clear ; however , the magnetic impurity in fe@xmath2cosib@xmath1 with @xmath4 near 725 k is likely associated with the unidentified reflections in the x - ray diffraction data for this material ( fig .
[ fig : pxrd]e ) based on diffraction and magnetization measurements on samples of varying purity . the curie temperature of fe@xmath0sib@xmath1 could not be determined because it is above 780 k , the maximum temperature for these measurements .
the downturn seen at the highest temperatures indicates @xmath4 is near but larger than this value .
this is consistent with the literature report of @xmath4 = 784 k for fe@xmath0sib@xmath1 , although some change in @xmath4 may be expected to occur with variation in si / b or p / b stoichiometry in these compounds @xcite .
measurements on fe@xmath0pb@xmath1 ( fig .
[ fig : jvst]a ) give @xmath4 = 640 k , similar to the reported values of 615@xmath14639 k @xcite and 655 k @xcite .
substitution of 20% mn for fe has a small effect on the determined @xmath4 values . in fe@xmath0pb@xmath1 @xmath4
is increased by 10 k , while in fe@xmath0sib@xmath1 it is decreased by 14 k. these variations are within the limits that might be expected for p - b or si - b antisite defects , based on the the report for fe@xmath0p@xmath15b@xmath16 @xcite .
a more dramatic change is observed for co substitution .
replacing one of five fe atoms with co reduces @xmath4 by more than 100 k in both fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 .
magnetization vs applied magnetic field measured at 300 k for the p - containing compounds ( a ) and the si - containing compounds ( b ) .
magnetization normalized by the saturation value is shown in ( c ) to facilitate comparisons of anisotropy fields .
note @xmath17 , and divide @xmath18 in tesla by @xmath19 to convert to _ m _ in emu/@xmath20 , or by @xmath21 to convert to _ m _ in a / m .
, width=312 ] results of magnetization measurements at 300k are shown in figure [ fig : jvsh ] .
all materials show ferromagnetic behavior .
the saturation magnetization ( @xmath22 ) , defined here as the value measured at an applied field of 5 t , for each material is listed in table [ tab : properties ] .
determinations of @xmath22 were also performed at 5 k and the results are also listed . a low temperature extrapolation value of 1.73 @xmath5/fe has been reported for the magnetic moment in fe@xmath0pb@xmath1 @xcite .
recently lamichhane _ et al .
_ reported the saturation magnetization of this compound to be 1.72 @xmath5/fe at 2 k and about 910 ka / m ( 1.1 t ) at 300 k @xcite .
the present measurements ( table [ tab : properties ] ) are in reasonable agreement with these reports . in general
, the magnetization is higher in the si - based compounds than in the analogous p - based compounds .
this is consistent with room temperature mssbauer spectroscopy studies which find larger hyperfine fields at the fe nuclei in fe@xmath0sib@xmath1 ( 17@xmath1423 t ) than in fe@xmath0ps@xmath1 ( 15@xmath1421 t ) @xcite .
substitution of co strongly reduces the magnetic moment , as seen in figure [ fig : jvsh ] and table [ tab : properties ] . comparing @xmath22 at 5 k , a reduction of 16% and 20%
are observed when 1/5 of the fe is replaced by co in fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 , respectively .
assuming that the average magnetic moment on fe remains the same when co is substituted gives an average ordered moment per co of 0.31 @xmath5 in fe@xmath2copb@xmath1 and 0.00 @xmath5 in fe@xmath2cosib@xmath1 .
the absence of a large ordered moment on co is consistent with our preliminary measurements , which suggest the absence of ferromagnetism in co@xmath0pb@xmath1 .
substitution of mn has a smaller effect on the magnetization . in both fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 , substituting 1/5 of the fe with mn
reduces @xmath22 by 6% at 5 k. this suggests that mn does have an ordered moment in these compounds .
indeed , the pure mn analogues are known to be ferromagnetic @xcite . assuming no change in the average moment per fe atom occurs with mn substitution , the data in table [ tab : properties ] gives 0.85 @xmath5 and 1.27 @xmath5 per mn at 5 k in fe@xmath2mnpb@xmath1 and fe@xmath2mnsib@xmath1 , respectively .
the conclusion that substituted mn is ferromagnetic while substituted co is not may explain the sharp decrease in @xmath4 that accompanies the addition of co but not mn ( fig .
[ fig : jvst ] and table [ tab : properties ] ) .
the rate at which the magnetization approaches saturation ( fig .
[ fig : jvsh ] ) is determined by the magnetic anisotropy .
this can be parameterized by the anisotropy field @xmath23 , which is the applied field required to saturate the material in the magnetically hard direction .
since a polycrystalline material with randomly oriented crystallites contains some fraction with the hard axis along the measurement direction , _ j _ vs _ h _ will also saturate near the anisotropy field , but the approach is typically quite gradual and @xmath23 can be difficult to define precisely in this way . in some cases ,
the singular point detection technique @xcite can be used to identify anomalies in d@xmath24/d@xmath25 which occur at @xmath23 , but that was not successful using the data for the present samples .
this technique requires that there is a sharp ( ideally discontinuous ) change in the slope of @xmath13(@xmath26 ) at @xmath23 in the hard direction , so that the second derivative contains a local minimum at this field .
but this minimum occurs on top of a smoothly changing background arising from the portions of the sample with the hard axis aligned in other directions .
better estimates can typically be obtained from measurements on single crystals or magnetically aligned powders .
this was recently done for fe@xmath0pb@xmath1 single crystals @xcite , and @xmath27 was determined to be 0.38 mj / m@xmath28 at room temperature . the anisotropy field can then be estimated by @xmath29 to be 0.83 t. an anisotropy field closer to 1 t might be estimated from the data shown in fig .
[ fig : jvsh]c , and this disagreement might be due to the neglect of demagnetization effects in the present analysis . with these limitations in mind , the data in fig .
[ fig : jvsh]c suggest that co and mn substitution reduces @xmath23 in fe@xmath0pb@xmath1 , but may slightly increase @xmath23 in fe@xmath0sib@xmath1 .
measurements on single crystals would be desirable to confirm this .
first principles calculations of fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 were performed to attain theoretical insight and understanding of the magnetic properties of these materials , using the plane - wave density functional theory code wien2k @xcite .
we used the generalized gradient approximation of perdew , burke and ernzerhof @xcite . for the p compound we used sphere radii of 1.77 , 2.15 and 2.06
bohr for b , fe and p and an rk@xmath30 of 7.0 was used . here
rk@xmath30 is the product of the smallest sphere radius and the largest plane wave vector . for the si compound we used sphere radii of 1.73 , 2.24 and 1.82 bohr for b , fe and si , respectively , with an rk@xmath30 of 7.0 .
approximately 1000 k - points in the full brillouin zone were used for these calculations , with spin - orbit coupling included ( excepting the internal coordinate optimization ) .
we briefly summarize the results for the p compound ( a more detailed discussion of these calculations can be found in ref . ) .
we find an average saturation moment of 1.79 @xmath31/fe , in good agreement with our experimental value ; this value includes small negative moments ( @xmath32 ) on the p and b atoms and a small orbital fe moment of approximately 0.03 @xmath31/fe . from the calculations we find uniaxial anisotropy , as in the experiment , with a magnetic anisotropy constant k@xmath33 of 0.46 mj / m@xmath34 , which yields an anisotropy field h@xmath35 of approximately 0.9 tesla , in reasonable agreement with the experiment .
the situation for the fe@xmath0sib@xmath1 compound is significantly more complex and interesting . as with the p compound
we find a ferromagnetic state with a saturation moment of 1.83 @xmath31/fe , including negative moments of -0.06 @xmath31 on the si and -0.1 @xmath31 on the b and a small orbital moment of 0.04 @xmath31 per fe .
this result compares well with the measured value of 9.06 @xmath5 per formula unit or 1.81 @xmath5 per fe(table [ tab : properties ] ) .
we present the calculated density - of - states in fig .
[ fig : dos ] below , which depicts a substantial exchange splitting of 2 - 3 ev as well as an electronic structure dominated around the fermi level by the fe , as expected in an fe - based ferromagnetic material . the calculated density - of - states of fe@xmath0sib@xmath1,width=312 ] for the calculations of magnetic anisotropy we used higher precision , with as many as 10,000 k - points in the full brillouin zone used in the self - consistent runs . unlike in the p compound @xcite , however , we find _ planar _ anisotropy with a first anisotropy constant k@xmath33 of -0.42 mj / m@xmath34 .
although fe@xmath0sib@xmath1 is a uniaxial magnet at room temperature , a spin reorientation occurs below room temperature as noted above and first observed in ref .
, so that our calculation is consistent with the ground state _
= 0 configuration .
note that the theoretical calculations were performed using a 300 k structure ( with internal coordinates optimized ) , so that the observed spin - reorientation does not appear to be the result of lattice expansion but may rather originate in dynamical effects .
these two scenarios were hypothesized @xcite as competing explanations for the spin reorientation @xcite in the manganese - based ferromagnet mnbi , which is planar below 90 k and uniaxial above .
here we suggest that dynamical effects from the variation in magnetic anisotropy associated with the motion of the ions away from their equilibrium positions are responsible for the spin - reorientation .
this is an effect distinct from both that of lattice expansion as well as that of the general reduction of magnetic moment with increasing temperature , which generally reduces magnetic anisotropy .
we have also performed calculations of the magnetic properties ( using the virtual crystal approximation ) of the partially mn substituted compound fe@xmath2mnsib@xmath1 . as with the other compounds we find a strongly ferromagnetic ordering , with average spin moment 1.88 @xmath5 per transition metal , which is slightly larger than the experimental value of 1.50 @xmath5 per transition metal ( table [ tab : properties ] ) .
it is possible that some of the alloyed mn anti - align with the fe , which would reduce the overall moment . unlike in the pure compound fe@xmath0sib@xmath1
we find _ uniaxial _ anisotropy in fe@xmath2mnsib@xmath1 .
the calculated anisotropy constant k@xmath33 is 0.26 mj / m@xmath34 , yielding an anisotropy field of 0.47 t , in reasonable agreement with the experimental results from fig .
5 . while a microscopic explanation for the change in anisotropy with mn alloying is not readily available , we note that similar effects have been calculated to occur in other fe - containing compounds such as fe@xmath36sn@xcite .
thus , the present result should not be considered anomalous or unusual , and may motivate further study of mn substitution as a way to manipulate the magnetic anisotropy in other fe compounds .
we find the tetragonal cr@xmath0b@xmath3 structure - type for all compositions and temperatures studied .
anisotropic thermal expansion is observed in fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 , with a relative change in _ a _ being 4@xmath146 times larger than that of @xmath11 between 300 and 20 k. fe@xmath0sib@xmath1 in fact displays negative thermal expansion along the @xmath11-axis over this temperature range .
all of the compounds have the easy axis of magnetization along the crystallographic _
c_-axis at room temperature .
the saturation moment of fe@xmath0sib@xmath1 is measured to be 1.81 @xmath5/fe at 5 k , in good agreement with density functional theory calculations which find 1.83 @xmath5/fe .
inspection of isothermal magnetization curves at 300 k suggest that fe@xmath0sib@xmath1 has a smaller magnetic anisotropy than fe@xmath0pb@xmath1 .
calculations ( _ t _ = 0 ) predict planar anisotropy for fe@xmath0sib@xmath1 , in agreement with experimental findings @xcite , of similar magnitude but opposite sign from that found in fe@xmath0pb@xmath1 @xcite . the substitution of co and mn does not strongly change the strength of the anisotropy near room temperature , but may slightly decrease it in fe@xmath0pb@xmath1 and slightly increase it in fe@xmath0sib@xmath1 .
an anomaly in the magnetization associated with the rotation of the spins into the _ ab_-plane is observed near 170 k in fe@xmath0sib@xmath1 .
no such anomaly is seen in fe@xmath0pb@xmath1 , fe@xmath2cosib@xmath1 , or fe@xmath2mnsib@xmath1 down to 30 k , showing that the substitution of co or mn for fe suppresses the spin reorientation .
our calculations agree with this finding , showing uniaxial anisotropy in fe@xmath2mnsib@xmath1 .
substitution of co is seen to strongly suppress the ferromagnetic behavior , reducing the magnetic moment by about 20% and the curie temperature by more than 100 k. mn substitution has a much smaller effect .
this is consistent with co being non - magnetic ( which is non - intuitive ) and mn coupling mostly ferromagnetically with the fe in these structures .
the results presented here demonstrate chemical tuning of the strength and anisotropy of the ferromagnetism in fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 , and that first principles calculations can reliably predict the magnetic moments and easy axis of magnetization .
therefore we expect that continued work involving closely coupled experiment and theory efforts may identify compositional modifications that would lead to enhanced permanent magnet properties in this family of compounds and others .
experimental work ( m.a.m . ) was supported by the u. s. department of energy , office of energy efficiency and renewable energy , vehicle technologies office , as part of the propulsion materials program
. theoretical work ( d.s.p . ) was supported by the critical materials institute , an energy innovation hub funded by the us department of energy , energy efficiency and renewable energy , advanced manufacturing office .
23ifxundefined [ 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 10.1109/tmag.2011.2166975 [ * * , ( ) ] link:\doibase 10.3891/acta.chem.scand.13 - 0433 [ * * , ( ) ] link:\doibase 10.3891/acta.chem.scand.14 - 1403 [ * * , ( ) ] link:\doibase 10.3891/acta.chem.scand.16 - 0001 [ * * , ( ) ] < go to isi>://wos : a1967a494600002 [ * * , ( ) ] link:\doibase 10.1016/0022 - 4596(75)90084 - 5 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1088/0031 - 8949/17/2/005 [ * * , ( ) ] @noop `` , '' link:\doibase 10.1016/j.jmmm.2009.03.067 [ * * , ( ) ] link:\doibase 10.1063/1.3518064 [ * * ( ) , 10.1063/1.3518064 ] @noop link:\doibase 10.1088/0031 - 8949/17/1/008 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | crystallographic and magnetic properties of fe@xmath0pb@xmath1 , fe@xmath2copb@xmath1 , fe@xmath2mnpb@xmath1 , fe@xmath0sib@xmath1 , fe@xmath2cosib@xmath1 , and fe@xmath2mnsib@xmath1 are reported .
all adopt the tetragonal cr@xmath0b@xmath3 structure - type and are ferromagnetic at room temperature with easy axis of magnetization along the _ c_-axis . the spin reorientation in fe@xmath0sib@xmath1
is observed as an anomaly in the magnetization near 170 k , and is suppressed by substitution of co or mn for fe .
the silicides are found to generally have larger magnetic moments than the phosphides , but the data suggests smaller magnetic anisotropy in the silicides .
cobalt substitution reduces the curie temperatures by more than 100 k and ordered magnetic moments by 16 - 20% , while manganese substitution has a much smaller effect .
this suggests mn moments align ferromagnetically with the fe and that co does not have an ordered moment in these structures .
anisotropic thermal expansion is observed in fe@xmath0pb@xmath1 and fe@xmath0sib@xmath1 , with negative thermal expansion seen along the _ c_-axis of fe@xmath0sib@xmath1 .
first principles calculations of the magnetic properties of fe@xmath0sib@xmath1 and fe@xmath2mnsib@xmath1 are reported .
the results , including the magnetic moment and anisotropy , and are in good agreement with experiment . |
the understanding of the proton - neutron pair - correlations is one of the major goals of the spectroscopy of nuclei with @xmath4 .
the progress in sensitivity achieved with the large @xmath5-ray detector arrays combined with mass seperators permits us to do detailed spectroscopy in the mass 80 and 50 regions .
the advent of radioactive beams will hopefully allow us to study even heavier @xmath2 nuclei .
these experimental opportunities revived the theoretical activities devoted to the study of the proton - neutron pairing .
what are the consequences of proton - neutron pairing for the excitation spectra ? this is an important question , because the excitation spectra and in particular rotational bands are the information that come from @xmath5 spectroscopy . in this lecture
we shall address a more specific question : what are the consequences of the * proton - neutron pair - field * for the rotational spectra ?
the restriction to the pair - field has the advantage that its symmetries show up directly in the spectra , which are qualitatively different for the various symmetry types .
the pair - field appears when the mean - field approximation is applied , which results in the hartree - fock - bogolubov ( hfb ) equations @xmath6 where @xmath7 the quantities in the brackets @xmath8 in ( [ hfbg ] ) and ( [ hfbd ] ) are the antisymmetric uncoupled matrix elements of the interaction . in eq .
( [ hfbh ] ) , we have introduced the isospin label @xmath9 for neutrons and protons , respectively and rearranged the chemical potentials @xmath10 and @xmath11 which constrain @xmath12 and @xmath13 into @xmath14 and @xmath15 which fix mass @xmath16 and the isospin projection @xmath17 , respectively . the hfb solutions are obtained by solving the equations ( [ hfb ] ) -([hfbt ] ) self - consistently .
the pairs of states @xmath18 that define the pair - field ( [ hfbd ] ) can be rewritten in a coupled representation as @xmath19 , which explicitly indicates the isospin @xmath20 and @xmath21 and @xmath22 denote all quantum numbers except the isospin . if @xmath1 , the pair - field is an isoscalar and for @xmath23 it is an isovector .
the proton - proton ( pp ) pair - field has @xmath24 and the neutron - neutron ( nn ) has @xmath25 .
there are two proton - neutron ( pn ) pair - fields with @xmath26 and @xmath27 .
we use the lower case letters @xmath20 and @xmath21 for the isospin of the pair - field in order to avoid confusion with the isospin of the states , which we denote by @xmath28 and @xmath17 .
let us restrict to the simple case of one @xmath29-shell , for which the symmetries become most obvious .
the generalization to many @xmath29-shells is straight forward .
then the field is specified by @xmath30 , where @xmath31 label the angular momentum of the pair .
anti - symmetry implies that for @xmath0 the angular momentum @xmath32 is even and for @xmath1 it is odd .
the monopole @xmath33 and the dipole @xmath34 are the most important pair - fields because they have the largest matrix elements for a short range interaction as well as a sufficient number of states available to built up correlations .
field has also a large matrix element but there is only one such state in the @xmath29-shell .
it may built up correlations in the space of many @xmath29-shells .
since its symmetries are the same as for the @xmath35 field , we shall not consider it further . ] in the following we assume that the pair - field is either @xmath36 or @xmath37 .
this is the most common solution of the hfb equations . but
coexistence of @xmath0 and @xmath1 fields is possible , as demonstrated by a. goodman @xcite .
we shall proceed with the assumption that for @xmath38 the @xmath2 nuclei have a @xmath0 pair - field at low spin .
this assumption is supported by the experiments with two - particle transfer reactions on these nuclei . as reviewed by bes , broglia , hansen and nathan @xcite
, the observed cross section can be very convincingly be interpreted in terms of a @xmath0 pair - field .
a dynamic picture of the field is used and multi - boson excitations are considered .
we shall stay within the static hfb approach and consider the consequences of the @xmath0 pair - field for the energy spectra .
the following discussion is based on the material in ref.@xcite .
the pair - potential @xmath39 has the same symmetries as the pairing - tensor @xmath40 , because the interaction is an invariant with respect to the symmetry operations . for @xmath0 monopole pairing
we have @xmath41 it is a vector in isospace in spherical representation .
the pair - potential @xmath42 can be written in an analogous way .
the original two - body hamiltonian conserves the isospin ( we neglect the coulomb interaction ) .
it is a scalar with respect to rotations in isospace .
we consider the case @xmath43
. then @xmath44 .
the mean field routhian @xmath45 ( cf .
( [ hfbh ] ) ) consists of isoscalar terms , except the pair - potential , which is an isovector .
therefore , it breaks the isospin symmetry spontaneously before discussing the breaking of the symmetry by the isovector pair - field , it is useful to state the familiar case of spontaneous breaking of the spatial isotropy by a mean - field solution with a deformed density distribution ( c.f .
ref.@xcite ) .
since the two - body hamiltonian is isotropic , this symmetry is broken spontaneously .
there is a family of mean - field solutions with the same energy which correspond to different orientations of the density distribution .
all represent one and the same intrinsic quasiparticle configuration , which is not an eigenfunction of the total angular momentum .
any of these solutions can be chosen as the intrinsic state .
the principal axes of its density distribution define the body - fixed coordinate system .
the states of good angular - momentum are superpositions of these states of different orientation , the weight being given by the wigner @xmath46-functions .
thus , the relative importance of the different orientations is fixed by restoring the angular - momentum symmetry . at the simplest level of the cranking model , which is valid for sufficiently strong symmetry breaking ,
the energy of the good angular - momentum state is given by the mean - field value .
let us now consider a @xmath0 hfb solution found for the @xmath43 system .
the @xmath0 pair - field @xmath42 is a vector that points in a certain direction in isospace , breaking the isospin symmetry .
since the two - body hamiltonian is isospin invariant , the symmetry is a spontaneously broken and all orientations of the isovector pair - field : @xmath47 are equivalent .
[ delta.fig ] illustrates this family of hfb solutions , the energy of which does not depend on the orientation of the pair - field . in particular
, the cases of a pure pn- field ( z - axis ) and pure pp- and nn - pair - fields ( y - axis ) represent the same intrinsic state .
hence , at the mean - field level the ratio between the strengths of pp- , nn- and pn- pair - fields is given by the orientation of the pair - field , which is not determined by the hfb procedure .
the relative strengths of the three types of pair - correlations becomes only definite when the isospin symmetry is restored .
the symmetry breaking by the isovector pair - field has been discussed before in @xcite , where references to earlier work can be found .
the most simple way to restore the symmetry is the above discussed rotor limit , which assumes that the rotation is slow as compared with the intrinsic motion . for the isospace it was discussed in @xcite . in essence
it amounts to pick one of the orientations of @xmath48 and call this the intrinsic state .
the direction of the y - axis , corresponding to @xmath49 , is particular useful , as will be discussed below .
the symmetry conserving wavefunction is a product of this intrinsic state and wigner @xmath46-function , which is the probability amplitude of the different orientations of the intrinsic state in isospin . for @xmath50 states it is a constant .
all orientations of @xmath42 are equally probable , corresponding to an equal amount of pn- , pp- and nn- correlation energy . in this way
the pn- pair - field reappears via restoration of the isospin symmetry , although the intrinsic state has only the pp- andd nn- pair - fields . for states @xmath51 ,
the @xmath46-function becomes more and more peaked in the y - direction , i. e. with increasing @xmath28 the orientations with @xmath52 become less and less probable .
this simple pair - rotational scheme is completely analogous to the familiar rotational bands .
the intrinsic excitations represent the @xmath50 states .
the energies of states with larger isospin are @xmath53 the moment of inertia for the isorotation can be calculated by means of the cranking procedure .
one solves the hfb eqs . for a finite
`` frequency '' @xmath54 and calculates @xmath55 the moment of inertia is approximately proportional to the level density at the fermi surface .
realistic interactions or shell model potentials are tuned such that the experimental level density is well reproduced .
hence , the pair - rotational energy , which is a combination of symmetry and the wigner terms of the binding energy , is expected to be reproduced well .
like in the case of spatial rotation , the intrinsic excitations are constructed from the quasi particles ( qps ) belonging to _ one _ of the orientations of pair - field .
we choose the y - direction , @xmath56 .
this is a particularly convenient choice because it permits to reduce the construction of the qp excitation spectrum to the familiar case with no pn - pairing @xcite .
the choice of the qp operators is not unique @xcite .
we choose them to be pure quasi neutrons or quasi protons and denote their creation operators by @xmath57 , where @xmath21 indicates the isospin projection .
they are pairwise degenerate , i.e. the qp routhians @xmath58 are equal .
our choice of the orientation of the intrinsic state has the advantage that its symmetries become obvious . since @xmath59 = [ e^{-i\pi n},{\cal h } ' ]
= 0,\ ] ] proton and neutron number parities are conserved .
that is states with even or odd @xmath12 are different and states with even or odd @xmath13 are different .
the hfb vacuum state has @xmath12 and @xmath13 even : configurations with an odd or even number of quasi neutrons belong to the odd or even @xmath12 , respectively , and the same holds for the protons .
in particular , the lowest @xmath50 state in odd - odd @xmath43 nuclei is a two - qp excitation and different from the ground state of its even - even neighbor , which is the vacuum .
however , not all qp configurations are permitted .
if @xmath44 , the qp routhian commutes with @xmath60 @xmath61 = 0\ ] ] this implies that the qp configurations have @xmath60 as a good quantum number . since @xmath62 ,
only configurations with @xmath63 are permitted .
the detailed discussion of this restriction can be found in @xcite . as a study case
, we used the deformed shell model hamiltonian which consists of a cranked deformed one - body term , @xmath64 and a scalar two - body delta - interaction @xcite .
the one - body term is the familiar cranked - nilsson mean - field potential which takes into account of the long - range part of the nucleon - nucleon interaction .
the residual short - range interaction is specified by the delta - interaction , @xmath65 where , @xmath66 we use @xmath67 as our energy unit and the deformation energy @xmath40 and is related to the deformation parameter @xmath68 .
we have diagonalized the hamiltonian ( [ h ] ) exactly for neutrons and protons in the @xmath69 shell , for which @xmath40=1.75 approximately corresponds to @xmath70 .
in addition to its invariance with respect to rotations in isospace ( 1 ) is invariant with respect to @xmath71 , a spatial rotation about the x - axis by an angle of @xmath72 . as a consequence ,
the signature @xmath73 is a good quantum number @xcite , which implies that the shell model solutions represent states with the angular momentum @xmath74 , @xmath75 integer .
the exact energies obtained by diagonalizing the shell model hamiltonian ( [ h ] ) for @xmath76 particles in the @xmath69 shell are shown in the upper panel of figs .
[ e33.fig ] .
the states are classified with respect to the isospin and the signature .
we have solved the hfb equations self - consistently for ( 4 + 4 ) at @xmath77 .
the solution is a @xmath0 pair - field . in order to construct mean field solutions for finite frequency we adopted the cranked shell model ( csm ) approach @xcite .
the fields @xmath78 and @xmath79 determined for @xmath77 are kept fixed for all other values of @xmath80 .
they are also used to describe the ( 3 + 3 ) and ( 3 + 4 ) systems , for which only @xmath81 and @xmath54 are adjusted to have @xmath82 and @xmath83 at @xmath77 .
earlier studies of a small number of particles in a @xmath29-shell showed that the csm gives better agreement with the exact shell model than than demanding full self - consistency for all @xmath80 @xcite .
[ qp.fig ] shows the quasiparticle routhians @xmath84 .
all are two - fold degenerated .
they correspond to a quasi proton and a quasi neutron , which are labeled , respectively , by a , b , c , ... and a , b , c , ... , adopting the popular csm letter convention .
the configurations are constructed by the standard qp occupation scheme , as described in ref .
the vacuum [ 0 ] corresponds to all negative qp orbitals filled .
it has has signature @xmath85 , even @xmath12 and @xmath13 and @xmath63 .
it represents the even - spin @xmath50 yrast band of the @xmath86 system .
the ab - crossing at @xmath87 corresponds to the _ simultaneous _ alignment of a proton- and a neutron- pair ( because the routhians are degenerate ) .
[ jx.fig ] demonstrates that the csm approximation describes the double alignment fairly well .
it also shows a shell model calculation where we took off all the @xmath1 components of the delta interaction .
the crossing appears at almost the same frequency as in the calculation with the full interaction .
hence the possible @xmath1 correlations can not influence the crossing in an important way , as has been speculated @xcite .
the lowest two - qp excitation is generated by putting one quasi proton and one quasi neutron on the lowest routhian .
we denote this configuration by @xmath88_0 $ ] .
it has @xmath63 and thus correspond to a @xmath50 band .
the subscript indicates the isospin @xmath28 of the configuration .
the total signature is @xmath89 and corresponds to an odd - spin band .
the particle numbers @xmath12 and @xmath13 must be odd , because exciting one quasi neutron changes @xmath12 from even to odd or from odd to even and the same holds for the quasi protons .
thus @xmath88 $ ] is the lowest @xmath50 odd - spin band in the odd - odd @xmath43 system .
[ e33.fig ] shows the csm estimate for this band .
the configuration @xmath90 $ ] is the second odd - spin @xmath50 band and @xmath91 $ ] the first even - spin @xmath50 band in the odd - odd system .
the configuration @xmath92 $ ] does not generate a new state , because only the superposition @xmath93-[a , b])/\sqrt{2}$ ] corresponds @xmath63 , the other must be discarded . to keep the notation simple ,
we label the configuration as @xmath91 $ ] .
but it is understood that the superposition is meant .
the lower panel of fig . [ e33.fig ] shows these , which represent the three lowest @xmath50 bands .
the lowest @xmath94 band is pair - rotational level based intrinsic vacuum state , which we denote by @xmath95_1 $ ] .
its energy is given by ( [ epairot ] ) and the @xmath96 is found by `` cranking in isospace '' according to ( [ jpairot ] ) . the comparison with the shell model calculation in fig .
[ e33.fig ] demonstrates that the simple pair - rotational scheme reproduces well the position of the @xmath94 even - spin band relative to the three lowest @xmath50 bands , the relative position of which is also well reproduced by the csm .
the appearance of the @xmath94 even - spin band below the @xmath50 bands is a specific feature of the @xmath97 system .
( in odd - odd nuclei with @xmath98 all bands start with an energy larger than @xmath99 . )
its low energy for @xmath77 has the consequence that the @xmath94 even - spin band is crossed by the aligned odd - spin @xmath50 band .
this crossing has been observed in @xmath100rb @xcite .
the similar energy of the @xmath94 and @xmath50 states at @xmath77 appears as a cancellation between the pair - gap and the isorotational energy .
the configuration @xmath101_0 $ ] is shifted by @xmath99 with respect to the qp vacuum @xmath95_0 $ ] .
the configuration @xmath95_1 $ ] is shifted by @xmath102 .
both quantities are nearly equal .
this is not a special feature of the @xmath29-shell model , as discussed now .
the energy difference between the lowest @xmath50 and @xmath94 states in odd - odd @xmath43 nuclei has recently by studied by p. vogel @xcite and a. macchiavelli _
et al._@xcite .
it turns out to be few 100@xmath103 for all nuclei with @xmath104 , except @xmath105 and 46 .
the small difference is an indication for the presence of the @xmath0 pair - field : the @xmath50 state lies at @xmath106 because it is a two - qp excitation .
the @xmath94 state lies at @xmath107 because it is the zero - qp state but the first excited state of the pair - rotational band . since the two terms are about the same the two states have almost equal energy .
one may derive experimental values for @xmath39 from the odd - even mass differences and for @xmath107 from the experimental energies @xmath108 within a isobaric chain .
these independently determined values are consistent with the experimental energy differences @xmath109 observed in the odd - odd nuclei @xcite . for the even - even @xmath43
nuclei the @xmath50 state is the vacuum .
then lowest @xmath94 state must be a two - qp excitation , which lies at @xmath110 , which is about 5@xmath111 .
the experimental energy of the lowest @xmath94 state agrees well with this estimate @xcite .
the fact that @xmath112 holds not only the experimental values but also for the single @xmath29-shell model points to a general feature , which remains to be understood . since we
the orientation of @xmath42 with @xmath113 is a legitimate choice for the intrinsic state , the analysis of rotational bands in realistic nuclei can be carried out along the familair scheme without a pn- pair - field .
one has only to take into account the possibility of low lying pair - rotational states and the exclusion of states due to the condition @xmath63 .
this sheds light on the results of the recent analyzes of high spin data in nuclei with @xmath114 and 1 by means of this conventional approach @xcite , which find good agreement between theory and experiment .
for @xmath114 the first excited pair - rotational state has @xmath115 .
it lies at least @xmath116 times higher than in the nuclei with @xmath117 , where it has @xmath94 . in the @xmath118 nuclei
it lies even higher .
thus the lowest bands are only the intrinsic excitations , which can be described as qp excitations with @xmath113 .
these results support our suggestion that in the investigated nuclei with @xmath119 there is strong @xmath0 pairing . in this connection
we want to point out once more that the fact that the mean field theory without an explicit pn- pair - field works well does by no means imply that there is no @xmath0 pn- field . on the contrary , it must have a strength comparable with the pp- and nn- fields in order to restore the isospin symmetry . fig . [ rb74.fig ]
displays the spectrum of @xmath120rb@xmath121 .
the upper panel also shows the @xmath118 bands measured in @xmath122kr@xmath123 .
they are isobaric analog to the @xmath94 bands in @xmath100rb and should give a good estimates of these bands .
since the @xmath94 states belong to an isobaric triplets , we set the ground state energy of @xmath100kr equal to the energy of the @xmath124 state in @xmath100rb .
[ qp72.fig ] shows the quasi neutron routhians for @xmath125 , which are nearly identical with the quasi proton ones .
the standard letter coding is used to label the qp routhians .
the use of a , a , ... indicates that the diagram is relevant for both neutrons and protons . in the lower panel of fig .
[ rb74.fig ] the rotational spectrum of @xmath120rb@xmath121 is constructed from the qp routhians .
the lowest @xmath50 configurations are generated by exciting a quasi proton and a quasi neutron .
the first is the positive parity odd spin band @xmath101_0 $ ] .
next , @xmath126_0 $ ] and @xmath127_0 $ ] are expected . as discussed in section
[ t=0.sec ] , the condition @xmath63 permits only one linear combination of the two excitations , obtained by exchanging the quasi proton with the quasi neutron , which we arbitrary label by only one of the terms in order to keep the notation simple .
the lowest @xmath94 bands are generated by cranking in isospace , using the realistic deformed nilsson potential ( for details see @xcite ) .
the lowest band is the vacuum @xmath95_1 $ ] .
it is crossed by the @xmath50 band @xmath101_0 $ ] , which has a large alignment .
the crossing frequency is fairly well reproduced .
thus it seems , that this crossing is a phenomenon belonging to the realm of @xmath0 pair - correlations .
the csm assumptions of fixed deformation and pairing are too inaccurate for the high frequency region .
of course one may combine the concept isorotation with a more sophisticated mean - field calculations .
[ rb74trs.fig ] presents the spectrum of @xmath100rb as an example . only pp- and nn- pairing is considered , but in addition to the monopole a quadrupole pair - field is taken into account . for each configuration and frequency @xmath80
, the deformation parameters are individually optimized .
the calculations of @xcite for the yrast sequence in @xmath100kr are used for the configuration @xmath95_1 $ ] and the results of an analogous trs calculation @xcite for @xmath100rb are used for the configurations @xmath101_0 $ ] and @xmath127_0 $ ] .
the relative energy of the @xmath50 and @xmath94 bands is calculated by setting at @xmath77 the energy difference between the configurations @xmath95 $ ] in @xmath128 and @xmath129 equal to the experimental value for the isorotational energy . the same harris reference as used for the experimental routhians
is subtracted from the calculated ones .
the calculated spectrum now agrees rather well with the data at high @xmath80 .
since the @xmath1 pair - field is an isoscalar @xmath130=0,\ ] ] and the qp routhian conserves the isospin .
the qp operators have @xmath131 and either @xmath132 ( neutron + proton hole ) or -1/2 ( proton + neutron hole ) .
the qp vacuum has @xmath50 .
the one - qp excitations have @xmath133 .
the two - qp excitations can be combined into @xmath134 states , analogous to eq .
( [ t1field ] ) , and into @xmath50 states , which correspond to the odd linear combination of the @xmath117 pairs .
the field conserves the parity of the total number of particles @xmath135 this means that even-@xmath16 nuclei and odd-@xmath16 nuclei correspond to different qp ( even or odd number of ) and have different excitation spectra .
however , it does not conserve the neutron or proton number parity separately @xmath136 this means that the same qp with zero or an even number of excited qps represents both even - even and odd - odd nuclei .
the wave function is a linear composition of states of even - even and odd - odd particle numbers .
adjacent even - even and odd - odd @xmath43 nuclei should have similar excitation spectra . in order to understand this statement better ,
let us briefly return to the familiar case of the pure neutron pair - field . for a with an even number of qps
the wave function is a linear combination of states with even particle number .
it may be viewed as the intrinsic state of a pair - rotational band .
the different members of the band correspond to @xmath137 . in the limit of very strong breaking of the particle number @xmath12 conservation ,
all the members of the band have the same excitation spectrum , which is given by the different intrinsic states .
this situation is reached for a very strong pair - field , when @xmath39 is much larger than the distance between the single - particle levels . in reality , @xmath39 is only somewhat larger than the level distance ( about 3 times in the heavy nuclei ) and the pair - correlations do not completely smear out the region near the fermi surface .
still one observes quite a remarkable similarity between the spectra of the adjacent even- even nuclei .
for example , the first excited two - qp state appears always at about @xmath99 . for the @xmath1 pair - field
the situation is analogous .
however , adding proton - neutron pairs , brings us from an even - even nucleus to an odd - odd one and the again to an even - even one , etc .
this means , if @xmath39 is larger than the distance between the single - particle levels , the spectra of neighboring even - even nucleus and odd - odd @xmath43 nuclei must be similar .
this represents a clear evidence , which can be checked experimentally .
the @xmath1 field consists of @xmath35 pairs , which have the three projections @xmath138 , 0 , and 1 .
let us assume that @xmath139 is the projection on the x - axis , which is the axis of rotation .
the components have different signature @xmath140 let us consider the important case that the deformed potential @xmath78 ( cf .
( [ hfbg ] ) ) conserves the signature .
if there is a @xmath141 field the qp routhian conserves the signature if there are @xmath142 fields the signature is not conserved .
let us first consider the case @xmath141 .
it was first discussed by goswami and kisslinger @xcite .
it corresponds to the ( @xmath143 ) pairs in goodman s classification ( cf .
his lecture and @xcite ) .
he finds for the even - even @xmath43 nuclei with @xmath144 a @xmath0 pair - field at low spin .
this is in accordance with the our discussion of the spectra in section [ t0 ] .
the observed spectra of adjacent even - even and odd - odd nuclei up to mass 80 are distinctly different .
this excludes a @xmath1 pair - field with @xmath39 larger than the single particle level distance . as discussed before , the different spectra of the even - even and odd - odd @xmath43 nuclei can easily be understood in assuming a @xmath0 pair - field .
goodman finds a @xmath1 pair - field of the ( @xmath145)-type for @xmath146pd @xcite .
the potential @xmath78 is near spherical .
the chemical potential is situated in the @xmath147 shell , which is almost degenerate .
in such a situation one can expect that the spectra of the even - even and odd - odd neighbors are similar .
it would be interesting to see if the experiment confirms this .
goswami and kisslinger @xcite considered the possibility that @xmath39 is much smaller than the level distance @xmath148 .
in such a case , the gap between the ground state and the first excited state is much smaller in the odd - odd nucleus ( 2@xmath39 ) than in the even - even ( 2@xmath149 ) .
it is difficult to derive information about such a weak pair - field from the spectra .
similarly , it seems hard to extract from the spectra clearcut evidence for a mixed pair - field composed of @xmath0 and @xmath1 components , which goodman obtains for one set of input parameters
@xcite .
this type of pair - field is favored by rotation , because it carries .
one may speculate that such a field appears at high spin , where the rotation has destroyed the @xmath0 pairing and made the phase space available for the new type of correlation .
the @xmath150 pair - field has a special symmetry .
it is odd under @xmath151 ( cf .
( [ comn ] ) ) and under @xmath71 ( cf .
( [ coms ] ) ) . therefore it is even under the combination of both and the qp routhian is invariant , @xmath152 there is an analogy to reflection asymmetric axial nuclei @xcite . although both @xmath71 and the space inversion @xmath153 do not leave the qp routhian invariant , the combination @xmath154 does , which implies the quantum number simplex .
the bands are @xmath155 sequences of alternating parity .
the simplex determines which parity belongs to @xmath156 . in the same way @xmath157 implies the quantum number @xmath5 @xmath158 which takes the values 0 and 1 for even @xmath16 and @xmath159 for odd @xmath16 . refering to the analogy with the simplex we suggest the name gauge - simplex or shorter gaugeplex
it relates the parity of the neutron number ( or proton number ) with the signature @xmath73 , @xmath160 here we used @xmath161 .
a strong @xmath150 pair - field implies that adjacent even - even and odd - odd nuclei join into a pair - rotational band of fixed gaugeplex .
this means that the even-@xmath156 states become similar to the odd-@xmath156 states of the neighbor and vice versa . in order to investigate this interesting structure we carried out shell model calculations for our study case of a @xmath69-shell . in order to enhance the @xmath1 correlations we modified the interaction .
only the odd-@xmath32 multipoles of the @xmath162-interaction , which have @xmath1 , were taken into account ( @xmath1 @xmath162 -interaction )
[ pn44.fig ] and [ pn33.fig ] show the results .
it is seen that for @xmath163 the spectra of the systems @xmath164 and @xmath165 become very similar if the states of opposite signature are compared .
the similarity may be ascribed to a gradual built up of the @xmath166 correlations caused by the increasing frequency .
this interpretation is supported by the development of a gap between the yrast and yrare state . at @xmath167 , the distances between the lowest routhians are in units of @xmath168 : 2.4 , 1.0 , 1.2 for @xmath165 and 2.2 , 0.9 , 0.9 for @xmath164 .
so far we have not been able to find a self - consistent pair - field in this frequency region .
it is not clear at this point what this means .
the work is very recent
. it may be that we just did not use the right initial wavefunction for the iterative solution of the hfb equations .
it is possible that the pair - correlations are of vibrational type and there is no static hfb solution .
but it could also be that there is a completely different explanation for the similarity .
the spectra of @xmath169 nuclei in the region @xmath38 are consistent with a @xmath0 pair - field at low and moderate spin ( cf .
s. lenzi s lecture at this meeting ) .
this conclusion is in accordance with the analysis of two - particle transfer data , which has already provided strong independent evidence for this type of pair - field @xcite .
the @xmath0 pair - field breaks the isotropy with respect to rotations in isospace .
therefore , the mean field solutions must be interpreted as intrinsic states of pair - rotational bands .
this has two consequences . as intrinsic state
, one may use the orientation in isospace with a zero proton - neutron pair - field .
the intrinsic spectrum looks as if there was no such field . only for @xmath43 certain excitations ( @xmath170 )
are forbidden .
the proton - neutron pair - correlations appear via the restoration of the isospin in the total wavefunction . on top of each intrinsic state , there is a pair - rotational band .
however , these additional states with @xmath171 lie high in energy . only in the odd - odd @xmath43 nuclei
the first excited pair - rotational state based on the intrinsic ground state has a similar energy as the first excited intrinsic state .
hence the appearance of this low lying @xmath94 rotational band is a consequence of the @xmath0 pair - correlations . the @xmath1 pair - field , which has other symmetries than the @xmath0 field , leads to a different pattern of excited states . since the pairs carry finite ( @xmath35 is expected to be most important ) one must distinguish between the fields with signature 0 and 1 . a substantial pair - field with signature 0
would show up as similar spectra in adjacent even - even and odd - odd @xmath43 nuclei .
such a similarity is not seen in the experimental spectra .
it should appear around @xmath172 , where goodman @xcite predicts this type of pair - field at low spin .
the pair - field with signature 1 is favored by rotation .
it may appear at high spin .
a substantial field of this type would show up as similar spectra in adjacent @xmath43 nuclei , provided the odd / even spins in even - even nucleus are compared with the even / odd spins in the odd - odd nucleus .
the work was supported by the grant de - fg02 - 95er40934 . | the symmetries of the @xmath0 and @xmath1 pair - fields are different .
the consequences for rotational spectra are discussed . for @xmath0 ,
the concept of spontaneous breaking and subsequent restoration of the isospin symmetry turns out to be important .
it permits us to describe the proton - neutron pair - correlation within the conventional frame of pairing between like particles .
the experimental data are consistent with the presence of a @xmath0 field at low spin in @xmath2 nuclei . for a substantial @xmath1 field ,
the spectra of even - even and odd - odd @xmath2 nuclei become similar .
the possibility of a rotationally induced @xmath3 pair - field at high spin is considered .
2j^(2 ) |
next - to - leading - order ( nlo ) effects in electroweak interactions play a crucial role in tests of the standard model , and require careful theoretical evaluation . an excellent place to search for new physics , the deviation of the weak charge of the nucleon from its standard model prediction also requires considerable experimental and theoretical input .
the importance of theoretical predictions for the weak charge of the nucleon was recognized more that two decades ago by marciano and sirlin @xcite .
their original analysis was followed recently by significant theoretical work on the proton weak charge @xcite , which presented an updated standard model prediction for the weak charge of the proton , and estimated the qcd structure uncertainties .
computer packages such as _ feynarts _ @xcite , _ formcalc _
@xcite , _ looptools _ @xcite , @xcite and _ form _ @xcite have created the possibility to go one step further .
now , we have an option of calculating parity - violating nlo effects including all of the possible loop contributions within a given model .
we have already tested the automated approach to calculate one - quark radiative corrections @xcite . in the work presented here ,
we adopt _ feynarts _ and _ formcalc _ for the nlo symbolic calculations of amplitude or differential cross section in electron - nucleon scattering . using dirac and pauli - type couplings with the monopole form factor approximation
, we construct the computational model enabling _ feynarts _ and _ formcalc _ to deal with electron - nucleon scattering up to the nlo level .
since the weak charge of the nucleon is directly related to the form factors of the parity - violating part of the amplitude at the zero momentum transfer , we choose the calculation of the weak charge of the nucleon to be a suitable test our method . within the uncertainty , which mostly comes from the uncertainty of the current electromagnetic form factor measurements , our results agree with experiment .
we start with expressions for pauli and dirac couplings in terms of fermion weak and electric charges , and the definition and classification of one - loop radiative corrections .
after that , an example for the @xmath0 box type of correction for electron - proton scattering is considered and a computational model is proposed .
we proceed with computational details for the self - energy graphs and vertex correction graphs .
since we reserve full kinematic dependence in all types of our nlo calculations , it will make it easier to adopt our results to the current and future parity - violating experiments .
in the approximation where the nucleon behaves as a point - like particle , vector boson couplings obey general rules of electroweak theory . for left and right handed fermions , we can use the following structure for the @xmath1 type couplings : @xmath2,\label{e1.1}\ ] ] where @xmath3 are chirality projectors and @xmath4 represents the coupling strength for the left and right handed fermions , respectively .
substitution of @xmath5 into eq.([e1.1 ] ) yields the vector and axial vector representation of the coupling @xmath6 : @xmath7.\label{e1.2}\ ] ] couplings of vector bosons to fermions derived from the neutral current part of the electroweak lagrangian predict tree - level values of @xmath8 and @xmath9 given by : @xmath10 here , @xmath11 and @xmath12 are twice the value of the isospin and electric charge respectively , and @xmath13 and @xmath14 are @xmath15 and @xmath16 , where @xmath17is a weinberg mixing angle . in the case
when a photon couples to the nucleon , @xmath18 and @xmath19 .
to accommodate nucleon structure , we have the couplings preserve their vector and vector - axial structure , but with the charges replaced by the corresponding form factors .
the electromagnetic @xmath20 coupling has two vector components responsible for static electric and magnetic interactions : @xmath21,\label{e1.5}\ ] ] where @xmath22 and @xmath23 are the dirac and pauli form factors , respectively , and @xmath24 is the four - momentum transferred to the nucleon . as for @xmath25 , we have : @xmath26,\label{e1.6}\ ] ] with @xmath27 , @xmath28 and @xmath29 as weak dirac , pauli , and axial - vector form factors . according to the first line of the eq.([e1.3 ] ) , form factors @xmath27 and @xmath28 are expressed as : @xmath30 with @xmath31 .
for @xmath29 we have @xmath32 where @xmath33 is an axial form factor . to considerably simplify analytical expressions ,
in our computational model we use the monopole structure for form factors expressed as @xmath34 which is a reasonable approximation in our case .
the value of the parameter @xmath35 is found after the fit of the electromagnetic form factors by monopole approximation in the low momentum transfer region .
further details on how couplings where implemented in the computational model are given in the appendix . in order to define the next - to - leading - order hadronic corrections
, we are using the electron - nucleon parity violating hamiltonian in the form proposed by @xcite : @xmath36.\label{i1.4}\ ] ] form factors @xmath37 and @xmath38 represent perturbative expansion resulting in @xmath39 the superscript in @xmath40 represents the order of the perturbation ( `` zero''- tree level , `` one '' - one loop level ( nlo ) and so on ) . here
@xmath41 can be defined as an one - loop contribution to the parity - violating form factor , normalized by the fermi constant , @xmath42 .
a calculation of nlo form factors is related to the calculations of loop integrals represented by three topological classes : box , self energy , and vertex ( triangle ) graphs . to preserve gauge invariance we have to include all the possible bosons of the standard model in these topological classes .
taking into account that in the thooft - feynman gauge the contribution coming from the higgs scalar and gauge fixing fields is negligible , we choose to consider boxes and nucleon vertex corrections ( triangles ) with @xmath43 and @xmath44 vector bosons only .
for the rest of the graphs self - energies and lepton vertex corrections
we accounted for all the possible particles in the standard model .
accordingly , we consider nlo corrections for every class .
the precise formulae for an entire set of graphs are cumbersome , and it is not feasible to show them in the present work . here , we provide some details for @xmath0 boxes only .
our complete calculations are shown in @xcite .
the full set of diagrams applicable for this case can be found in @xcite . according to the feynman rules ( see , for example @xcite ) , the amplitude for a @xmath0 box ( see fig.([f1.1 ] ) )
can be written as : box diagram in the electron - nucleon scattering .
[ f1.1 ] ] @xmath46 in order to integrate the amplitude in eq.([e1.4 ] ) and the rest of the nlo loop integrals using _ feynarts _ and _ formcalc _ , it is necessary first to create a model file with declarations of all the fields participating in interaction and second to describe all the possible couplings of the fields on both generic and particle levels . for the case of electron - nucleon scattering , the only new particles in the model besides the particles of the standard model are neutron and proton , so it is rather straightforward to declare them in the _ feynarts _ model file . on other hand , the description of the couplings according to the eq.([e1.5 ] ) and eq.([e1.6 ] ) in the _ feynarts _ represents a certain problem . because we use the monopole form factor approximation ( see eq.([e1.9 ] ) )
, it introduces momentum dependence in the coupling s denominator . for _ feynarts _ to deal with momentum - dependent couplings with momentum dependence introduced in the denominator , these couplings have to be described as propagators . of course , a propagator introduced in the coupling conflicts with the declaration of the fields in the _ feynarts _ model files which use the same propagator notation .
we can solve the problem by transferring the monopole form factor from the coupling part of model to the part where fields are described using propagator notation . for boxes that can be achieved by starting with the general definition of a four - point tensor integral of rank @xmath47 in the form @xmath48 by adding the monopole form factor approximation into the above definition
, we obtain a six - point tensor integral which can be reduced into a combination of four - point integrals by using the following expansion : @xmath49 here @xmath50 represents the last two terms of the product in eq.([e1.19 ] ) multiplied by the two monopole form factors of the box diagram .
explicitly , the right - hand side of eq.([e1.20 ] ) can be written in the form@xmath51 substitution of the last equation into amplitude eq.([e1.4 ] ) results in an expansion where each term of the sum represents an amplitude constructed for electron - nucleon scattering with the point - like nucleon .
also , in this consideration , the couplings between nucleon and vector bosons are multiplied by the factor @xmath52 , and the structure of the second , third and fourth terms of eq.([e1.20a ] ) suggests the introduction of fictitious vector boson particles @xmath53 with fixed masses @xmath54 .
these bosons have no physical meaning , of course , but they allow us to remove momentum dependence in the denominator of the coupling .
now we can model couplings between nucleon and vector fields using eq.([e1.5 ] ) and eq.([e1.6 ] ) with a monopole form factor moved into the definition of propagators for the new vector fields such as @xmath53 where @xmath55 . a diagrammatic representation of the proposed expansion
is given by the set of feynman graphs in fig.([fhdr1.3 ] ) .
box in terms of @xmath56 particles[fhdr1.3 ] ] as for the boxes , we have all we need to complete the automated construction of amplitudes in the _ feynarts _ and the calculations in the _
formcalc_. we should note that eq.([e1.4 ] ) is written for the @xmath0 box diagram only . to have a complete analysis , it is necessary to consider @xmath57 , @xmath0 and @xmath57 crossed boxes as well .
@xmath58 and @xmath59 box diagrams should be considered , too .
our calculations include 36 boxes in total . in total , 116 self - energy graphs and 6 counterterms contribute to the pv @xmath60 amplitude .
this includes gauge and the gauge - fixing fields , the higgs field , and virtual leptonic and quark pairs in creation - annihilation processes in the loops .
moreover , the vertex @xmath61 does not belong to the loop integrals and plays the role of a multiplicative factor proportional to the coupling defined in eq.([e1.5 ] ) and eq.([e1.6 ] ) . in _
the self - energy loop integrals can be evaluated using the expansion given in the fig.([fhdr1.4 ] ) , and then tensor decomposition and tensor reduction techniques applied , leaving the final result as a combination of one and two point scalar integrals which we compute using gauss integration subroutines of _
looptools_. partners . here ,
factor @xmath52 used for the couplings of the box diagrams should be replaced by one.[fhdr1.4 ] ] to cancel ultraviolet divergences we employ an on - shell renormalization scheme according to @xcite .
we have to assume that quarks in the self - energy loops are free , but it places certain constraints . since quarks are confined , a qcd strong quark - quark interaction should always be considered .
it is possible to bypass these complications by replacing quarks with pions , or use `` free '' quarks but with adjusted effective masses . here
, we use the second approach with the effective mass of the quarks coming from a fit of hadronic vacuum polarization to the measurements of qed cross section of the process @xmath62hadrons normalized to the qed @xmath63 cross section .
the real part of the renormalized hadronic vacuum polarization satisfies the dispersion relation : @xmath64 with @xmath65 being a very well known experimental quantity and used here as an input parameter . ]
hadronic vacuum polarization @xmath66 is related to the truncated @xmath67 renormalized self - energy ( see fig.([hdrvp ] ) ) by the following expression : @xmath68 which can be easily evaluated by employing the free quark approximation .
an updated value of the dispersion integral , along with a logarithmic parametrization , can be taken from @xcite .
a new reported value coming from the light quark contribution at @xmath69 is @xmath70 .
this value can be reproduced by eq.([e1.33 ] ) using the following masses of the light quarks : @xmath71 ( corresponds to @xmath72 ) .
clearly , the values of the light quark masses at low - q scattering processes should be adjusted by using this approach , but with @xmath73 calculated in the region of @xmath74 gev .
the simple logarithmic parametrization can be used here to extract quark masses at low momentum transfer : @xmath75 with @xmath76 and @xmath77 parameters taken from @xcite . for low - momentum transfer experiments , the c.m.s .
energy is @xmath74 @xmath78 , which gives , @xmath79 .
the vertex correction contributions can be split into two classes . in the first class , where the electron vertex is at the one - loop level
, the amplitude is calculated according to @xcite . as in the case of the self - energy graphs ,
the hadronic vertex does not belong to the loop integrals , and therefore the pv amplitude is constructed according to the sm feynman rules employed in the _ feynarts _ package .
generally , the electron vertex corrections have an infrared divergence at @xmath80 and are treated by the bremsstrahlung contribution considered in @xcite . the second class of the triangle graphs are hadronic vertex corrections . in this case , to evaluate the amplitude automatically using the packages _ feynarts _ and _ formcalc _ , we construct an expansion similar to that given for the box diagrams . to work out the set of rules for the triangle topology , it is sufficient to consider the example in fig.([vr.cr.z.ex ] ) . ] for the graph in fig.([vr.cr.z.ex ] ) , the amplitude denominator has the structure @xmath81 which can be easily expanded into @xmath82 here , the coefficients @xmath83 and @xmath84 are equal to @xmath52 respectively .
@xmath85 , and @xmath86 can be calculated using the following formulae : @xmath87 the expansion of the amplitude denominator in eq.([e1.35 ] ) has a simple graphical representation ( see fig.([hdr.vert.exp ] ) ) and suggests , in this particular case of the triangle topology , that a set of virtual particles @xmath88 and @xmath89 should be introduced in the nlo hadronic vertex corrections . and @xmath89 partners.[hdr.vert.exp ] ]
although ultraviolet divergences are absent in the hadronic vertex corrections due to the additional terms proportional to @xmath90 in the coupling , it is still necessary to compute the wave function renormalization with some details given in the appendix . as well as in the case of electron vertex corrections
, the nucleon vertex will have an infrared divergence at the pole @xmath80 ( for the proton ) .
a detailed discussion of the treatment of this type of divergence is given in @xcite .
the application of the methods described above for @xmath60 scattering can be found in calculations of the weak charges of the nuclei .
consider the parity - violating hamiltonian in eq.([i1.4 ] ) . here ,
for a heavy nucleus we have a coherent effect for @xmath91 : @xmath92 the contribution coming from @xmath93 is small as it depends on unpaired valence nucleons .
the latter determines the hamiltonian for the electron parity - violating interaction with the nucleus in the following form : @xmath94 a relation between the weak charge @xmath95 and form factors @xmath96 is straightforward : @xmath97 if we take into account only the leading order of the interaction , the weak charge of the proton and neutron have the simple definitions : @xmath98 and for the nucleus @xmath99 where @xmath100 @xmath101 are the weak charges of the proton and neutron including nlo corrections . as one can see from fig.([c1np.1 ] ) , the input of the nlo corrections is much more significant for the proton than for the neutron .
this makes experiments involving neutrons more interesting , as they allow for the cleaner weak charge extraction . and @xmath102 .
the dotted line represents parity - violating form factor at the tree - level ( lo ) only , and the solid line shows the result with nlo contribution included ( lo+nlo ) .
[ c1np.1],title="fig : " ] and @xmath102 .
the dotted line represents parity - violating form factor at the tree - level ( lo ) only , and the solid line shows the result with nlo contribution included ( lo+nlo ) .
[ c1np.1],title="fig : " ] also , form factors @xmath103 and @xmath104 normalized by the monopole term @xmath105 ( see fig.([c1np.1 ] ) ) do not exhibit strong momentum transfer dependence and hence are model independent .
this justifies to some extent our choice of the monopole form factor approximation .
extrapolation of the @xmath103 and @xmath104 to zero momentum transfer point ( see fig.([c1np.1 ] ) ) gives : @xmath106 and @xmath107 .
the most important of these numbers is the result for @xmath108 it has been evaluated most recently in @xcite , leading to @xmath109 .
the available results from atomic parity - violating experiments for the weak charges of @xmath110 and @xmath111 ( @xcite and references within , and @xcite ) can be used as an experimental test of the theoretical predictions : @xmath112 here , the errors are statistical , systematic and coming from an uncertainty of the atomic - physics theory , respectively . for example , in the case of @xmath113 one should observe @xmath114 parity - violating electric dipole transitions in order to extract the weak charge of @xmath115 this requires an accurate knowledge of the atomic wave functions .
some of the analysis of the issues involved , with an extensive reference list , is given in @xcite and @xcite .
using eq.([d1.3 ] ) and eq.([d1.5 ] ) we compute the following results for the corresponding nuclear weak charges : @xmath116 which clearly agree with the experiment .
the theoretical uncertainty is estimated quite generously and comes mostly from the numerical integration and the extrapolation to the zero momentum transfer point . although the nlo corrections contribute only @xmath117 to the results in eq.([d1.7a ] ) , their value is still about four times larger than our theoretical uncertainty . thus , the more precise is the experiment ( like @xmath118 ) , the more important it becomes to evaluate most carefully the full set of the applicable nlo corrections .
although our model predictions for the nuclear weak charges are in good agreement with the available experimental results , to allow more definitive conclusions about the validity of the proposed computational model , the experimental errors for precision measurements of the weak charge of the nucleon will have to be reduced . any significant deviation of the weak charge of the proton from the standard model prediction at low @xmath119 would be a signal of new physics . the proton s weak charge @xmath120 is a well - defined experimental observable . at @xmath121
the asymmetry can be parametrized as @xmath122[q^{2}q_{weak}^{p}+q^{4}b(q^{2})],\ ] ] where @xmath123 is a function of sachs electromagnetic form factors @xmath124 related to the dirac and pauli form factors by the following expression : @xmath125 the measurement of @xmath120 will be done by the @xmath95 collaboration @xcite , and may have extremely interesting physical implications .
see , for example , @xcite .
it will allow one to determine the proton s weak charge with @xmath1264% combined statistical and systematic errors , which leads to @xmath127 for the running of the weak mixing angle . the standard model evolution predicts a shift of @xmath128 at low @xmath119 with respect to the @xmath129 pole best fit value of @xmath130 the weak mixing angle at the energy scale close to the @xmath129 pole was measured very precisely , but a precision experimental study of the evolution of @xmath131 to lower energies still has to be carried out .
the asymmetry measurements proposed for @xmath95 experiment will go as low as @xmath132 , making it a very competitive experiment . using the dirac and pauli form factors
, we can now compute the extensive set of one - loop hadronic electroweak nlo corrections along with the weak charges of the proton and neutron . using a monopole approximation for the form factors , we modified general electroweak couplings by inserting the appropriate form factors into the vertices .
the monopole structure of the form factors allows us to substitute one feynman diagram with `` structured '' nucleon with a set of diagrams involving only point - like nucleon .
this expansion can be visualized by adding fictitious additional vector bosons of mass @xmath133 . for renormalization , we choose the on - shell renormalization scheme . in conclusion ,
it is evident that computational symbolic packages such as _ feynarts _ and _ formcalc _ can be efficiently adapted for the theoretical evaluation of nlo effects in electron - nucleon scattering .
refs @xcite and @xcite give some additional details .
a newly proposed measurement of the electron weak charge in parity - violating moller scattering at 12 gev at jlab could be a very interesting challenge , for example .
the authors thank malcolm butler of saint mary s university for his useful comments .
we are also grateful to shelley page of university of manitoba for some of her explanations on the @xmath95 experiment .
this work has been supported by nserc ( canada ) .
to introduce couplings between vector bosons and nucleon into the model files of _ feynarts , _ instead of vector and axial - vector representation , we employ use of chirality projectors .
that can be achieved if we compare eq.([e1.1 ] ) , eq.([e1.2 ] ) with eq.([e1.6 ] ) , combined with eq.([e1.7 ] ) and eq.([e1.8 ] ) , then it is possible to write @xmath134 & = \nonumber \\ \label{e1.11}\\=ie\left[g_{l}^{z - n}\left(q\right)\gamma_{\mu}\varpi_{-}+g_{r}^{z - n}\left(q\right)\gamma_{\mu}\varpi_{+}+\frac{i}{2m_{n}}\sigma_{\mu\alpha}q^{\alpha}f_{2}\left(q\right)\right]\nonumber \\
\nonumber \\\gamma_{\mu}^{\gamma - n}\left(q\right)=ie\left[\frac{g_{l}^{\gamma - n}\left(q\right)+g_{r}^{\gamma - n}\left(q\right)}{2}\gamma_{\mu}+\frac{i}{2m_{n}}\sigma_{\mu\alpha}q^{\alpha}f_{2}\left(q\right)\right]=\nonumber \\ \label{e1.12}\\=ie\left[g_{l}^{\gamma - n}\left(q\right)\gamma_{\mu}\varpi_{-}+g_{r}^{\gamma - n}\left(q\right)\gamma_{\mu}\varpi_{+}+\frac{i}{2m_{n}}\sigma_{\mu\alpha}q^{\alpha}f_{2}\left(q\right)\right ] , & & \nonumber \end{aligned}\ ] ] where @xmath135 in eqs.([e1.11 ] ) and ( [ e1.12 ] ) , we have adopted the general structure of the coupling from eq.([e1.1 ] ) , with coupling strengths @xmath136 and @xmath137 given by eq.([e1.13 ] ) .
vector part of the coupling @xmath138 also can be replaced by the chirality projectors @xmath139 in the following way@xmath140\varpi_{-}+\left[\gamma_{\mu},\not q\right]\varpi_{+}\right).\label{e1.13b}\end{aligned}\ ] ] moreover , to adopt _ feynarts _ representation of the coupling through the product of generic and class type of the couplings , we introduce the following matrix representation of eq.([e1.11 ] ) and eq.([e1.12 ] ) : @xmath141\varpi_{-}$ , & $ \left[\gamma_{\mu},\not q\right]\varpi_{+}$\end{tabular}\right)\overrightarrow{g}_{nnv}\left(q\right),\label{i1.1}\ ] ] with @xmath142 defines coupling between classes of the vector bosons and nucleons and expressed as a @xmath143 matrix @xmath144 the second column of @xmath142 represents counterterms of the first order .
the pauli form factor @xmath145 in eq.([i1.2 ] ) has the following structure : @xmath146 the coupling defined in eq.([i1.2 ] ) has a counterterm part at the one - loop level represented by the matrix , which has following structure : @xmath147\\
g_{r}^{v - n}re[\delta\ , f_{r}]\\ -\frac{1}{4m_{n}}f_{2}^{v - n}\left(q\right)re[\delta\ , f_{l}]\\ -\frac{1}{4m_{n}}f_{2}^{v - n}\left(q\right)re[\delta\ , f_{r}]\end{array}\right),\label{e1.37}\ ] ] where the hadronic field renormalization constants @xmath148 are computed using the expansion on fig.([hdr.nucl.ren.cons ] ) .
more details are given in ref.@xcite . | we show how computational symbolic packages such as _ feynarts , formcalc , form _ and _ looptools _ can be adopted for the evaluation of one - loop hadronic electroweak radiative corrections for electron - nucleon scattering and applied to calculations of the nucleon weak charge .
several numerical results are listed , and found to be in good agreement with the current experimental data . |
astrophysical foregrounds set an unavoidable limitation to precise measurements of primordial cosmic microwave background temperature fluctuations , ( @xmath3 ) , even at frequencies close to the intensity peak of the relic radiation . on the other hand , the millimeter region of the e.m .
spectrum corresponds to a minimum in the spectral energy distribution of the galaxy and of the majority of the extragalactic sources .
this minimum occurs at the cross over between radio and dust emission , which steeply rises with frequency , and it is only weakly dependent on the relative intensity of the two components ( see , e.g. , * ? ? ?
* ; * ? ? ?
therefore , this spectral region is the best one for mapping cmb anisotropies . as for extragalactic point sources
i.e. , galaxies seen as a ` point like ' object through the beam , being their typical projected angular size @xmath4 their isotropic distribution gives rise to a contaminating signal which presents the same _ average _ level all over the sky , at a given frequency @xcite .
thus , it can be only reduced by the identification and detection of as many sources as possible . on the other hand ,
the relatively large beam sizes and high flux detection limits of current and forthcoming earth , balloon and space borne experiments ( e.g. , boomerang , vsa , cbi , dasi , acbar , archeops , nasa wmap and esa _ planck _ missions ) imply that only few relatively bright sources can be detected and removed from current as well as future cmb sky maps ( see , e.g. , * ? ? ?
therefore , the contribution of the highly undetected extragalactic source populations to cmb temperature fluctuations must be accurately estimated to avoid an unwanted incorrect reconstruction of the angular power spectrum of primordial anisotropies .
this problem is particularly important at intermediate to high multipoles , i.e. @xmath5 , where intrinsic cmb anisotropies are damped .
many estimates of temperature fluctuations arising from a poisson distribution of extragalactic sources have been worked out in the past by @xcite , @xcite , @xcite , @xcite , @xcite .
soon after , a thorough analysis of the extragalactic foreground contributions to small scale fluctuations over the full wavelength range from @xmath6 cm to @xmath7 @xmath8 m , which improved on previous ones , has been presented by @xcite . assuming a poisson distribution of point sources in the sky , they found that the central frequency channels of the _ planck _ mission will be ` clean ' ( i.e , only a few high latitude pixels will be contaminated by bright undetected sources ) .
as for radio selected extragalactic sources , which contaminate cmb anisotropies at _ planck _ low frequency instrument ( lfi ) channels @xcite , their clustering signal was found to give a generally small contribution to temperature fluctuations , thanks to the broadness of the local luminosity function @xcite and of the redshift distribution of sources which dilute the clustering signal . at higher frequencies ,
the clustering of far ir selected dusty galaxies was found to give a more relevant albeit not dominant contribution to temperature anisotropies . on the other hand , recent results coming from the sub
millimeter common use bolometric array ( scuba , * ? ? ?
* ) surveys are giving increasing evidence that many of the sources detected in this frequency region are ultraluminous star
forming galaxies at @xmath9 ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a detailed evolution model which fully accounts for the scuba counts in the framework of hierarchical clustering scenarios has been recently presented by @xcite . in this model
, scuba sources mainly correspond to the phases of intense star formation in large spheroidal galaxies at substantial redshift , a process which should be almost completed at @xmath10 . as a consequence ,
the relevant redshift range is considerably limited and the dilution of the clustering signal results relatively reduced ; correspondingly , the amplitute of @xmath3 fluctuations due to clustering is expected to be large . as a preliminary estimate of this signal , have shown that if the sub mm ( dusty ) galaxies detected by scuba at 850 @xmath8 m do cluster like lyman - break galaxies @xcite , anisotropies due to clustered sources dominate the poisson ones at all angular scales for @xmath11 ghz .
more recently , various papers @xcite exploited the evolution model of @xcite to work out precise theoretical predictions on the power spectrum of cmb temperature anisotropies due to scuba selected galaxies . in all these works ,
the authors exploited the linear and non linear growth of galaxy clustering with redshift and the evolution of the bias factor @xcite .
all these recent studies clearly indicate that dusty galaxies at intermediate to high redshifts are strongly correlated in the sky and , thus , they are giving rise to a clustering signal which probably dominates over the poisson one in almost all _ planck _ high frequency instrument ( hfi ) channels @xcite . in view of the previously quoted results and , moreover , for having a tool which could help in the analysis of current as well as future all
sky maps , we decided to try a different approach for estimating the power spectrum of eps temperature fluctuations .
this approach is based on a fast algorithm capable of simulating all sky maps as well as sky patches on which eps are distributed following a given two
point angular correlation function , @xmath1 .
the method , thoroughly explained in sections 2 and 3 and also discussed , albeit partially , by @xcite , works as follows .
first of all , a simulated map of poisson distributed eps is created ( see , e.g. , @xcite ) .
the only basic requirements are : a ) an input model for the number counts ; b ) a fixed value for the pixel size . subsequently , to obtain a map of extragalactic sources whose positions are correlated in the sky , it is necessary to select the _ most reliable _ angular correlation function for that particular source population we want to simulate .
then , by working in the fourier space , we modify the poisson density field by introducing the chosen angular power spectrum , @xmath0 , as discussed in section 2 . finally , after having applied an inverse fourier transform to the map , the fluxes of the simulated sources are distributed in the sky by adopting the same model number counts used in the first step of the process to obtain the poisson density field ( see section 3 ) .
eventually , from this map it is easy to estimate the angular power spectrum of eps by applying the standard fourier analysis . from the very beginning , we focused on 2d sky maps thus avoiding the problem of 3d simulations , much more expensive in terms of cpu time for allowing us the creation of a lot of maps , under the most different assumptions , in relatively short cpu times . moreover , with the purpose of producing all sky maps which could be used by the cmb community in general and by the _
consortia , in particular , we adopted the standard healpix pixelization scheme of @xcite throughout the paper .
more specifically , we chose a pixel size of [email protected] arcmin@xmath13 , i.e. nside=2048 ( @xmath14 pixels ) , in the healpix scheme .
as for the flat sky patches , we use the standard angular dimension , @xmath15 deg@xmath13 , and a pixel size of [email protected] arcmin@xmath13 , unless otherwise stated . in this latter case ,
our simulated maps are thus divided in @xmath16 pixels .
the outline of the paper is as follows . in
2 we briefly discuss the basic formalism focussing on the assumptions adopted for efficiently simulating maps of clustered eps . 3 will be devoted to describe the model number counts and the angular correlations functions of extragalactic sources here adopted . in
4 we present our predictions on the power spectra of temperature fluctuations due to the clustering of ` known ' source populations .
moreover , we discuss the excess power at high multipole recently detected by the cbi and dasi experiments . finally , 5 summarizes our main conclusions . a flat cosmological model with @xmath17 km / s / mpc , @xmath18 is used throughout the paper .
anyway , we have to remind that our conclusions are only very weakly dependent on the underlying cosmology given that the number counts of extragalactic sources are mainly determined by the evolution with cosmic time of the number density and/or of the emission properties of the underlying source populations ( see , e.g. , * ? ? ?
* ; * ? ? ?
we remind here that all the predictions discussed in this paper are based on the assumption of ` point like ' sources .
as shown by @xcite , this is a good approximation as far as the angular sizes of sources do not exceed the beam width .
this is generally the case in our simulations , since the adopted pixel size
i.e. , the resolution element of the map is @xmath19 arcmin@xmath13 and there are very few ( very bright ) low redshift sources in the microwave sky which show angular sizes greater than @xmath6 arcmin .
moreover , the few bright extended sources will be anyway detected and removed by standard or new detection techniques @xcite . as a preliminary step towards the realization of a flat ( a sky patch ) or spherical ( all sky ) map of clustered sources , we distribute point sources by adopting a simple poisson distribution for the number , @xmath20 , of sources per pixel .
the mean of @xmath20 is then @xmath21 , i.e. the average number of sources per pixel , which is determined by the total number counts @xmath22 , where @xmath23 indicates the differential source counts and @xmath24 the flux limit for source detection .
] at a given frequency @xmath25 ( see 3 ) .
these sky maps have been already used and discussed in many other papers ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and the reader can refer to them for more details .
therefore , at this first step , the sources are not spatially correlated in the sky .
we then define the _ projected _ density contrast at a given point ( pixel ) as @xmath26 , being @xmath27 the average number of sources per pixel .
therefore , the covariance function of the projected density contrast is the usual two point angular correlation function ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) @xmath28 where @xmath29 is the angular separation in the sky between the two positions and the brackets indicate an ensemble average . as a following step ,
we calculate the fourier transform of the density contrast @xmath30 where @xmath31 is the angular size of the map we are actually using .
it is very easy to show that the angular power spectrum @xmath32 is the fourier transform of the angular correlation function @xcite . , then @xmath33 . ]
the angular power spectrum depends only on @xmath34 , being the field homogeneous and isotropic .
thus , the basic idea of our method is to obtain _ a density field determined by a given power spectrum _ or , equivalently , by the suitable angular correlation function .
we do this by calculating the fourier transform of the density contrast , @xmath35 , and obtaining its power spectrum , which is constant for all modes , @xmath36 , if sources are poisson distributed in the sky .
subsequently , we introduce the chosen angular power spectrum of correlated sources , @xmath37 , suitable for the particular source population adopted ( see 3.2 . ) , by applying the following formula @xmath38 then we apply the inverse fourier transform to @xmath39 which allows us the recovering of a new density field , @xmath40 , in which the poisson term has been modified by @xmath37 , i.e. the fourier transform of the chosen angular correlation function , @xmath1 . finally , we calculate @xmath41 which gives the modified number of point sources at each position in the map , according to the previously calculated new density field , @xmath40 . if @xmath42 in equation ( 3 ) we obtain , again , a pure poisson distribution
. different angular power spectra can be used for the simulations and they can be derived from the angular correlation @xmath1 , according to the formula of @xcite @xmath43 where @xmath44 is the zeroth - order bessel function . in this first approach
we have only taken into account the first two moments of the angular distribution , albeit this assumption can not give a full description of the statistical properties of the field .
anyway , as already pointed out by @xcite , this procedure proves `` safe '' given that the total number of bright sources and the number counts remain unchanged after having modified their spatial distribution by the suitable angular correlation function ( see also section 3.2.1 ) .
moreover , the recovered @xmath1 matches the input one quite well .
this result is displayed in figure 1 : in the upper panel we plot four @xmath37 , calculated by equation ( 5 ) , chosen as examples to test the procedure ( see caption ) . among them
, we use the well determined @xmath37 coming from the apm survey @xcite as the reference one for this test .
this angular power spectrum is better determined than the other ones due to the very great number of galaxies in the survey .
in fact , this @xmath37 is the only one which clearly shows the transition from the linear to the non linear regime . in all the other cases , the data are coming from much shallower surveys on a smaller sky area and , thus , they first converge to a constant value at low multipole . in the lower panel of figure 1
we show the recovered @xmath0 , calculated directly from only one all - sky map : the agreement is very good at all scales , with maximum relative differences @xmath45% .
the difference is slightly higher , @xmath46% , as expected , due to the higher sample variance . ] in this case we plotted the @xmath0 of a single map to show that the angular power spectrum can be well recovered from each map and it is not the result of averaging over @xmath47 simulations . in the same lower panel of figure 1
it is also plotted the average value and the @xmath48 confidence interval of the recovered @xmath0 , after 100 simulations ( in the case of a flat sky patch ) .
the result clearly indicate that the method works well : the slope of the power law @xmath49 is recovered with an error @xmath50% .
notice that the apm @xmath0 is the most difficult one to reconstruct : in all the other cases the reconstruction proves easier , since input @xmath0s are usually fitted by a simple power
law at all scales where they can be determined .
in the previous subsection we have shown how it is possible to distribute eps in the sky , by using their average number density per pixel , @xmath27 , which is determined only by the number counts at each frequency . in this way we have obtained a pure _ density field _ , characterized by the particular @xmath0 used in each map and by the average number , @xmath27 , of sources per pixel .
now , for converting it to a temperature map , @xmath51 , we have to distribute first the fluxes corresponding to the @xmath52 sources of the total counts .
then , the usual conversion @xmath53 is applied .
firstly , we have to know the differential counts at a given frequency and , in principle , for each source population which are giving us the number , @xmath23 , of extragalactic sources in each flux interval . at each frequency where it is estimated
, @xmath23 provides a specific distribution function for fluxes , which mainly depends on the cosmological evolution and on the emission properties of the underlying source populations .
therefore , it is necessary to find an efficient algorithm for distributing fluxes in the map and which satisfies two fundamental requirements : a ) the differential counts should be recovered within the statistical error determined by the sample variance ; b ) the `` input '' angular correlation function has also to be reconstructed , at least down to the flux limit of the sample by which that specific @xmath0 has been determined . as already pointed out by @xcite ,
this is a purely _ phenomenological _ approach aimed at simulating all sky maps of eps in a fast way and , at the same time , with the guarantee that the two previously quoted requirements are satisfied . in principle
, one can choose many different ways of distributing fluxes among pixels : e.g. , by simply distributing fluxes under the condition that in a pixel in which there are @xmath47 sources we have to put _ a corresponding number _
@xmath47 of fluxes , taken _ at random _ from the distribution given by the differential counts ; or , e.g. , by imposing that the brightest fluxes fall always in the highest density pixels , thus strengthening , like imposing a `` bias '' factor , the correlation function .
anyway , no matter the method of distributing fluxes you choose , the constraint will be always the same : to fulfil the above quoted , a ) and b ) , fundamental requirements .
to achieve this , and with the purpose of simulating maps of eps at cmb frequencies , we have first exploited the results coming from the analysis of the nrao vla sky survey ( nvss ) at 1.4 ghz discussed by ( hereafter bw02 ) . by analyzing the great amount of data of this large area survey , the authors are able not only to quantify the angular distribution of radio sources at this frequency but also to give estimates of the @xmath1 for different flux ranges .
therefore , by simulating a clustered distribution of eps at 1.4 ghz we can test the method more deeply : first of all , for the fact that at this frequency the differential counts are well constrained down to a few @xmath8jy and the @xcite(hereafter to98 ) model fits them quite well down to the faintest fluxes ; secondly , for having the possibility of testing the recovered @xmath1 , i.e. the corresponding @xmath37 , in different flux ranges . as a first approach , we applied the method of _ distributing the fluxes at random _ among pixels and the results
are shown in figure 2 . in each panel of the figure we plot the total @xmath54 , given by the poisson plus the clustering term ; we also represent the input and output @xmath37s ( converted to @xmath54s ) , due to the pure clustering term and for different flux limits , following table 1 of bw02 .
the power spectra are calculated from each map by averaging @xmath55 over the modes with the same @xmath56 .
the @xmath54s are then calculated from the @xmath0s just by multiplying them by the patch area , @xmath57 .
notice that when indicating a flux limit , @xmath24 , we do indicate that we have used _ only those pixels with resulting fluxes above _ the quoted @xmath24 for estimating the @xmath37 from the map .
this choice guarantees that we are correctly comparing the angular correlation function of the simulated map with the input one of bw02 .
the maps used for obtaining the results plotted in the _ left panels _ are constructed in the following way : a single map is obtained by summing up 4 different maps with the different amplitudes of @xmath1 given by bw02 corresponding to the flux ranges @xmath58 , @xmath59 , @xmath60 mjy and , finally , a map of poisson distributed sources in the range @xmath61 mjy , thus reaching the same @xmath62 on eps counts as the one used by @xcite . on the other hand ,
each final map from which the @xmath0s in the _ right panels _ of figure 2 . are calculated is obtained by only one map in which all sources with @xmath63 mjy are distributed in the sky according to the most accurate determination of @xmath1 given by bw02 ( see caption to figure 2 ) . notice that , _ in all cases _ , we first create the density field , as explained in 2.1 . , and then distribute _ at random _ the fluxes of the sources previously used for the corresponding density field . only sources ( and fluxes ) corresponding to the specific flux range used , at the chosen frequency ( 1.4 ghz ) , are thus distributed in each map . from figure 2
it is clear that : a ) the input @xmath37 , converted to @xmath54 , is well recovered from the simulated maps for every chosen @xmath24 in agreement with the original bw02 article , taking into account that @xmath37 is obtained by subtracting the poisson term from the total simulated @xmath0 ; b ) the poisson term gives rise to a much greater signal than the clustering term , in all the cases : the total power , @xmath64 , is two / three orders of magnitude above the clustering term .
moreover , the dependence on @xmath56 ( or @xmath65 ) of the poisson and of the clustering term is _ different _ : this well known result is a direct consequence of the increase of the clustering - to - poisson ratio at low multipoles @xcite .
c ) the reconstruction is performed with almost the same precision
average slopes of the @xmath37 : @xmath66 and @xmath67 , respectively by distributing sources using either different amplitudes , @xmath68 , for different flux ranges ( as in table 1 of bw02 ) or by using only one amplitude , the one quoted in the caption of figure 2 .
this is not a surprising outcome , given that bw02 state that _
`` ... the clustering amplitude is also independent of flux '' _ , like the slope of the power
law , at least in the flux ranges analyzed by them . due to this fact
, one could question our choice of distributing the faintest sources ( and , thus , fluxes ) following a poisson distribution and not by applying the same power
law used for the brightest ones .
actually , this choice is not relevant in view of simulating maps for low resolution experiments , like wmap and _ planck _ : the much more numerous fainter sources in the beam ( pixel ) dilute the clustering signal and only add some gaussian noise in the pixel ( see , e.g. , * ? ? ?
on the other hand , it should be taken into account in the case of simulating maps for the analysis of fields observed with much higher resolution , e.g. in the optical band .
the results plotted in figure 2 are quite encouraging and give us confidence in the method .
anyway , for completeness , we have also checked that if we _
force _ the fluxes of the brightest sources to fall in the higher density pixels
we clearly exceed the input value of the selected @xmath1 .
this is an expected result since all our `` input '' angular correlation functions have been determined , as usual , by _ flux selected _ samples of eps , and not by samples of objects defined by using some different selection criterion . on the other hand ,
if we want to reproduce the distribution in the sky of highly clustered source populations , like the dusty spheroids at high redshift , discussed in 3.2.2 , we have to force the brightest sources to fall in the highest density bins . for each particular source population
the correct way of distributing fluxes has to be found , with the constraint of being able to recover the corresponding observed / theorethical @xmath37 ( see 4.1 ) .
it is important to stress that the proposed method allows to add up _ as many source populations as needed_. however , it shall be generally sufficient to simulate only _ one source population _ , the one which dominate the counts in the flux interval of interest , except for some very specific frequency channel in which two , or more , source populations showing different clustering properties are giving a comparable contribution to the counts .
this is a great advantage , which reduces the total required cpu time , and is determined by the fact that the total variance of intensity fluctuations due to eps , @xmath69 , is obtained by adding up in quadrature all the contributions coming from different source populations .
, is given by @xmath70 , where the subindexes @xmath71 , @xmath72 and @xmath73 , indicate a specific source population , and the poisson and clustering terms , respectively , and @xmath74 with @xmath75 ( see e.g. , * ? ? ?
] therefore , the contribution of faint eps populations to the total @xmath76 is generally very small .
eventually , we remind that if we perform the fourier transform of the temperature map , @xmath77 , the temperatute power spectrum of the map is calculated by averaging over the modes with the same k values , @xmath78 . then , in the flat approximation
, @xmath54 can be estimated by @xmath79 , where is @xmath31 is the angular dimension of the sky patch .
as for all sky maps , the anisotropies of the cmb , @xmath80 , are usually expanded in spherical harmonics @xmath81 as usual , the angular power spectrum , @xmath54 , is then calculated averaging the coefficients @xmath82 with the same @xmath83 .
the coefficients @xmath84 are estimated by the routine _ anafast _ of the healpix package .
we have run our code in a sun blade 1000 workstation ultrasparc iii at 750 mhz and with 5 gbytes of ram memory .
the process of creating one all - sky map by distributing only one eps population takes from @xmath852 to @xmath8510 hours , depending on the total number of sources ( @xmath86-@xmath87 ) , for nside=2048 ( @xmath14 pixels ) in the healpix scheme .
the total time is determined by the total number of pixels , @xmath88 , and by the total number of sources , @xmath89 in the map : it goes basically as @xmath90 $ ] . the time required to the healpix routine _
`` synfast '' _ for simulating the pure density field at the same resolution is @xmath91 80 minutes .
the rest of the time is determined by the distribution of @xmath92 up to @xmath93 fluxes , depending on the frequency channel and on the source population . on the other hand , a 512@xmath12512 pixels flat patch of the sky
is simulated , by a 2.0 ghz processor with 512 mbytes of ram memory , in only a few seconds at 30 - 100 ghz and in @xmath94 minutes at 857 ghz , for one source population .
obviously , if we distribute in the sky more than one source population , adding up in flux the different simulated fields , the total time is correspondingly increased .
as a benchmark to perform our simulations , we have used the differential counts of the cosmological evolution model for radio selected sources by to98 which successfully account for the observed total number counts down to @xmath95 mjy and up to 8.44 ghz . moreover , also the relative ratio of `` flat''-to-``steep''-spectrum sources is well accounted for by the to98 model ( see , e.g. , @xcite ) .
recently , new results at 15.2 ghz @xcite suggested that the extrapolation of the to98 model counts overestimates their data by a factor of @xmath96 at the survey limit ( @xmath97 mjy ) . on the other hand ,
the recently published analysis of the foreground emission in the wilkinson microwave anisotropy probe ( wmap ) first year data @xcite has provided a full sky catalogue of 208 bright extragalactic sources with fluxes @xmath98 jy , of which only five objects could be spurious identifications .
the whole sample provides an average `` flat '' ( @xmath99 ) energy spectrum in full agreement with the assumptions of to98 , and number counts of bright sources at 41 ghz which appear to fall below the prediction of to98 at 44 ghz by a factor @xmath100 .
however , a direct calculation of the 33 ghz counts by the source catalogue in table 5 of @xcite gives 155 sources at @xmath101 jy on a sky area of 10.38 sr ( @xmath102 ) where the sample appears to be statistically complete . according to table 5 of @xcite .
moreover , by a re - binning of the data in the same table we found that the differential counts of wmap detected sources sink down below @xmath103 jy , suggesting the onset of incompleteness . ] with this latter source selection the average offset , @xmath104 , with the to98 model predictions is slightly reduced .
furthermore , two other recently published independent samples of extragalactic sources at 31 and 34 ghz from cbi @xcite and vsa @xcite experiments , respectively show that the to98 model _ correctly _ predicts number counts down to , at least , @xmath105 mjy ( cf .
13 of @xcite ) .
as pointed out by @xcite , it is also noticeable that the brightest extragalactic source in the wmap first year catalogue has a flux of @xmath106 jy at 41 ghz which , again , corresponds to the flux for which the to98 model predicts 1 source all over the sky .
another important result is that a good agreement is currently found between predictions based on the to98 model and wmap data on the excess angular power spectrum at small angular scales @xcite as well as on the angular bispectrum detected in the wmap q and v bands @xcite . on the other hand , a lot of new data on far
ir / sub mm source counts have piled up since 1998 , in particular by means of the scuba and mambo surveys .
these current data are better explained by new physical evolution models of far ir selected sources @xcite .
these models foresee a steeper slope of the differential number counts of eps than the to98 model do at intermediate fluxes , i.e. @xmath107 mjy , where the contribution of high redshift spheroids show up ( see , e.g. , * ? ? ?
in view of this , and for comparing our current estimates on the @xmath54 of eps with the theorethical predictions of @xcite , we have also used the model counts of @xcite at the frequencies of 217 and 353 ghz .
these are relevant frequencies , which correspond to the central channels of _ planck _ hfi , and where the contribution of dusty galaxies to cmb anisotropies should start to dominate over the one coming from radio selected sources . as for the `` input '' angular correlation function
, we have to use , obviously , the @xmath1 most appropriate to that specific source population we have to simulate ( see also 2.2 . ) . in this paper
we are mostly interested in simulating all sky maps at cmb frequencies in view of the optimization of current simulations and future data analysis for the _ planck _ mission .
in particular we would like to develop a useful tool for testing the efficiency of algorithms for source detection , for component separation , for detecting non gaussianity in the residual maps and , finally , for better denoising of current as well as of future maps . given the lack of determinations at cmb frequencies , we have , again , to rely on @xmath1s determined at frequencies close to the ones under study .
whereas in the optical domain the very deep large area surveys of the last decade have provided a lot of very precise determinations of the @xmath1
e.g. , apm , 2dfgrs and sdss sky surveys @xcite only a few determinations of the @xmath1 at radio ( cm ) frequencies are currently available . moreover , at hfi frequencies , the only data up to now available on clustering properties of eps are coming from the scuba sky fields .
these fields are of very small angular dimensions and , thus , no clear determination of the @xmath1 at these frequencies has been yet published . some theoretical analysis of the clustering properties of scuba galaxies
have been recently attempted @xcite .
for all the above reasons , and taking into account previous analyses on number counts of eps at cmb frequencies @xcite , we have made the following choices as for the most reliable @xmath1s to be applied in simulations at _ planck _ lfi and hfi frequencies .
all the most recent surveys of radio sources are confirming with a small offset the predictions on number counts made by to98 , at least up to @xmath108 ghz ( 3.1 . ) .
at these frequencies , the eps relevant at fluxes of interest for current as well as future cmb anisotropy experiments are mainly `` flat''spectrum sources , i.e. qsos , bl lacs and local agns , and , thus , the contributions coming from other fainter source populations can be neglected .
moreover , the number of `` inverted '' spectrum sources which could have represented a threat for cmb @xmath3 measurements is found to be always small @xcite .
therefore , we could be confident in adopting the angular correlation function determined by the parkes - mit - nrao survey at 5 ghz @xcite for simulating sky maps of clustered eps sources at _ planck _ lfi frequencies and angular resolutions . on the other hand , when analyzing cmb sky maps coming from high resolution experiments which probe fainter fluxes
one can rely on the @xmath1 estimated by from the nvss survey .
this @xmath1 refer to fainter sources than the @xcite one and represents a more realistic approximation to the clustering properties of cbi and bima radio sources ( see 4.2 ) .
in figure 3 , bottom panel , we plot the input differential counts of to98 model at 5 ghz , which match very well the observed ones ( as shown by figure 1 of to98 ) , and they are compared to the ones recovered from only _
one _ simulated map . distribution down to @xmath109 mjy .
] as it is apparent from figure 3 , the recovering of the differential counts from the simulated maps is achieved with very high precision down to very faint fluxes . in both cases , i.e. poisson and non - poisson ,
the counts recovered form the map start to exceed the input ones by a small factor only at fluxes @xmath110 mjy , whereas they fall rapidly below the input model counts at @xmath111 mjy . level ( 10 sources to be compared with 6 , in a flat sky patch ) . ]
source counts are increased at flux levels much fainter than the flux detection threshold , @xmath112 mjy , calculated at 5 ghz by using the same formalism as in @xcite , to98 for a top hat square beam of fwhm@xmath94 arcmin and without any other noise added .
therefore , the adopted method does not modify number counts at least down to the flux level at which eps can be detected with our resolution element and the clustering term , i.e. the adopted input @xmath1 , only negligibly increases the counts at faint fluxes , as expected in this frequency range .
level @xcite . ] in the case of poisson distributed sources ( empty circles ) the simulated counts fall below the input ones at @xmath113 mjy , the level at which the to98 model predicts @xmath851 source / pixel , with a pixel size @xmath114 arcmin@xmath13 . in the case of clustered sources , with the @xmath37 here
adopted , the limit of 1 source / pixel is reached at slightly brighter fluxes , i.e. @xmath115 mjy .
mjy is exactly the same as the one lost at fainter fluxes , due to the effect of source confusion with the fast convergence of the counts at @xmath116 mjy . ] in the top panel of figure 3 we plot the input and recovered @xmath37 of @xcite . at 5 ghz ,
given the shallowness of the survey , we can rely only on a preliminary estimate of @xmath117 for fluxes @xmath118 mjy and we can only apply our method by using a single @xmath1 . therefore , we have distributed eps in the sky in the same way as at 1.4 ghz but as in the case whose results are plotted in the _ right panels _ of figure 2 ( only one @xmath1 for all sources at @xmath63 mjy ) . as for the angular correlation function , the curves plotted in the top panel of figure 3 .
show , again , that the epss2d algorithm is capable of simulating maps of clustered eps in which the input @xmath37 is well recovered .
we chose to represent the result of 1000 simulations ( and the corresponding @xmath119 intervals ) , given that the reduced number of pixels with @xmath118 mjy implies that by only one realization of the sky the simulated @xmath0 results too noisy ( the poisson term always dominates , as clearly shown by the upper thick line ) . anyway , from 100 simulations only we recover an average slope of @xmath37 , @xmath120 and an amplitude whose offset with the input one is @xmath121% and always compatible at the @xmath122 level .
the other plotted curve ( see caption to figure 3 ) shows the effect of the dilution of the clustering signal , when lowering down the flux detection limit , as already discussed in 2.2 . for simulating sky maps of far
infrared ( far ir ) selected eps , which dominate the total number counts at @xmath123 ghz , we have relied on the new physical evolution model for sources recently proposed by @xcite .
this model , which successfully reproduces the observed counts and the available information on the redshift distribution of scuba sources , describes scuba sources as intermediate and high redshift proto - spheroidal galaxies mainly observed during their principal burst of star formation , whose evolution and duration is substantially affected by the growth and by the activity of a central super massive black hole and not only by supernova feedback . obviously , spheroidal galaxies ( spheroids ) are not the only source population contributing to number counts at far ir and sub millimeter wavebands
type galaxies , i.e. normal spiral , irregular and starburst galaxies , dominate the bright tail of number counts whereas spheroids provide , by far , the overwhelming contribution to @xmath23 in _ planck _ hfi channels for @xmath124 mjy . on the other hand ,
late type galaxies are relatively weakly clustered ( see , e.g. , * ? ? ?
* ) and , thus , their contribution to confusion noise turns out to be dominated by poisson fluctuations @xcite .
scuba sources are expected to show a strong spatial clustering , similar to the one measured for extremely red objects ( eros ; * ? ? ? * ; * ? ? ?
on the other hand , direct measurements of clustering properties of these sources is made difficult by the poor statistics .
however , @xcite found some evidence for clustering of the background source population in the hubble deep field observed at 850 @xmath8 m and tentative evidences of strong clustering , consistent with that found for eros , have been recently reported @xcite .
the power in excess over poisson fluctuations detected by @xcite is well accounted for by a two point angular correlation function of the form @xmath125 with @xmath126 in the range 12 arcsec .
as discussed by @xcite , the above value of @xmath126 is consistent with a number of data on clustering of scuba sources and can be well accounted for by a @xmath1 derived from physical assumptions on the evolution of clustering @xcite . in the framework of this physical evolution model for clustering , @xcite worked out a complete set of predictions on the angular power spectrum due to eps sources at hfi frequencies . given that we are currently interested in testing our algorithm for distributing clustered sources in cmb sky maps , in the following we adopt the @xmath1 of @xcite and the number counts of spheroids of @xcite for simulating all - sky maps of far ir selected eps .
therefore , from these maps we have to recover the same theoretical @xmath54 values of @xcite .
figure 4 and 5 display our current results on the temperature angular power spectrum , @xmath54 , due to clustered eps at the _ planck _
channels 30 , 100 , 217 and 353 ghz by applying the epss2d algorithm .
notice that each plotted @xmath54 has been calculated from only one all - sky map of eps , following the method explained in 2 and with the assumptions made in 3 .
however , for giving a confidence interval , we also plot the @xmath119 levels around the mean , having performed 100 simulations of flat sky patches . as in @xcite , to98 we represent the quantity @xmath127 ( in units of k ) .
as discussed in 3.2.1 , due to the relatively large beam area ( i.e. , bright detection limits ) of current as well as future all - sky cmb anisotropy experiments , we can be confident in simulating 2d all sky maps by using only `` flat''spectrum cm selected radio eps at @xmath128 ghz .
the angular power spectrum , @xmath37 , determined by using the angular correlation function of @xcite has been used , as before , given that the parent population of bright sources , i.e. blazars , is the same one as at 5 ghz .
we rely on this choice since current earth borne as well as satellite surveys are confirming that other source populations , i.e. inverted spectrum sources , are almost not sampled at bright fluxes if @xmath128 ghz @xcite . from figure 4
it is clear that the recovered @xmath54s of clustered eps are still compatible with the ones recovered from a poisson distribution of sources @xcite .
the contribution of clustered eps sources to temperature cmb fluctuations at these frequencies is _ found to be negligible at all angular scales _ , `` if sources are not subtracted down to fluxes well below the detection limit of the survey , thus greatly decresing the poisson fluctuations , while the contribution arising from clustering is only weakly affected '' @xcite .
in fact , if only bright sources at @xmath129 jy are subtracted out from the map
e.g. , the detection limit of the wmap survey
the @xmath54 of clustered eps still match very well the predictions of to98 .
these results are , again , in agreement with the well known observational result : `` the wide redshift range of radio sources washes out much of the clustering signal '' ( bw02 ) . from the same figure 4 it is possible to appreciate a small excess at the largest angular scales in the estimated @xmath54s , of clustered eps , if compared to the ones estimated by to98 .
this is in agreement with the well known outcome that the ratio of clustering to poisson fluctuations increases with increasing angular scale , i.e. at decreasing multipole number ( see , e.g. , * ? ? ?
* ; * ? ? ?
the @xmath54 of clustered dusty proto - spheroidal galaxies at 100 ghz ( see figure 4 ) confirms previous findings of @xcite on the confusion noise : the poisson term due to radio selected sources is always the dominant one at all multipoles at @xmath128 ghz . on the other hand , at @xmath130 ghz spheroids probably dominate the fluctuations due to eps.% .
as an example , at 100 and 143 ghz we find @xmath131 mjy , respectively , at the planck resolution , to be compared with the values @xmath132 mjy quoted by @xcite and by using the same detection thresholds .
the contribution of spheroidal galaxies is almost insensitive to the adopted flux limit , because of the very steep counts at bright fluxes . ] in figure 5 we plot our current outcomes at 217 and 353 ghz by using the eps counts of spheroidal galaxies of @xcite ( see 3.2.2 ) and by applying the @xmath1 discussed in @xcite . in this case ,
due to the very strong angular correlation function detected by scuba @xcite , for recovering the input @xmath133 , we have to force the brightest fluxes to fall in the highest density pixels . from figure 5
it is clear that a very good agreement is found between the recovered @xmath54s and the theoretical predictions of @xcite , at all multipoles .
the small scatter around the average value , displayed by the @xmath119 confidence intervals , shows that we can be confident in the @xmath54s recovered from each simulated map .
as previously noticed , at 217 ghz ( 3rd _ planck _ hfi channel ) a comparable contribution to bright eps counts is expected either from spheroidal galaxies and from radio selected `` flat''spectrum sources .
anyway , given that this latter population is much less clustered and , moreover , thanks to the dilution of the clustering signal due to their flat luminosity function and broad redshift distribution
the contribution of `` flat''spectrum eps falls below the one coming from spheroids ( see , e.g. , * ? ? ?
* ) . at 353 ghz ,
spheroids starts to dominate the counts at very bright fluxes and , thus , if they are strongly clustered as suggested by scuba surveys @xcite their @xmath54s are no doubt the dominant ones . as for other far
ir selected source populations , they give a small to negligible contribution to cmb temperature fluctuations , since they are much less clustered ( see 3.2.2 ; @xcite ) .
therefore , we can be sufficiently confident in our current results .
recent cmb experiments , aimed at probing arcmin angular scales , i.e. dasi , bima amd cbi , respectivey @xcite , have tentatively detected a signal at high multipoles well in excess on predictions for primordial cmb temperature anisotropies .
the interpretation of these outcomes is still open .
possible interpretations are in terms of the thermal sunyaev
zeldovich ( sz ) @xcite signal associated to the formation of spheroidal galaxies @xcite or in terms of the sz effect in galaxy clusters @xcite .
however , all these interpretations require very high values of the @xmath135 normalization . on the other hand ,
the emission from advection - dominated accretion flows ( adaf ) around supermassive black - holes in early - type galaxies is found to be not sufficient to explain the detected excess @xcite .
undetected eps can represent an alternative ( or , at least , a partial ) explanation to the excess .
given that extragalactic sources are , very probably , the dominant contributor to cmb fluctuactions at these small angular scales , they have to be carefully subtracted out . and ,
indeed , all the quoted works devoted a great effort to this subject . in figure 6 ( bottom panel )
we plot , as an example , the @xmath54 due to poisson distributed eps , as estimated by @xcite and our prediction at 30 ghz by using the to98 model counts and a detection limit @xmath136 mjy for sources .
the very good agreement between our current results and those estimates confirms , again , that the to98 model still holds well , down to fluxes of tens of mjy at @xmath137 ghz ( see section 3.1 ) .
moreover , this agreement also implies that source clustering _ can not _
add a relevant contribution to the observed signal , as previously discussed .
as discussed by @xcite , a quite substantial residual contribution of radio selected sources to the cbi signal is difficult to rule out , whereas the residual contamination of the bima data should be , very likely , smaller .
all published estimates of eps contamination of cmb maps at high multipoles have used a poisson distribution of extragalactic sources . by our simulation code and the to98 model counts we can now assess , with good confidence , the excess contribution due to clustered eps sources . in figure 6
we plot the total noise ( poisson+clustering ; continuous line ) and the pure clustering term ( dotted lines ) given by radio selected sources following the to98 source counts .
we model the clustering term by using the @xmath1 of @xcite and bw02 as discussed in sections 2 and 3 .
the top panel of figure 6 shows that undetected sources below @xmath138 mjy , the most recent detection limit quoted by @xcite , can easily explain the excess . if we guess that radio selected sources do cluster down to very faint flux limits ( @xmath109 mjy ) following the @xmath1 of @xcite , then the clustering term should give a non negligible contribution ( @xmath139-@xmath140k ) to the total fluctuations . on the other hand ,
if we adopt the @xmath1 of bw02 , which can represent a more realistic approximation to the clustering properties of faint undetected sources in the cbi fields , since we are dealing with very faint sources not sampled by the nrao - vla survey , the additional power due to clustering ( @xmath141-@xmath142k ) is indeed very small in comparison with the poisson term . as for the bima data plotted in figure 6
, our current estimates do indicate that poisson distributed and clustered eps selected at radio frequencies are the dominant contributor to the detected excess , by adopting the effective detection limit @xmath143jy , below which the counts of detected sources sink down @xcite , and by keeping the same assumptions as before for their clustering properties . on the other hand , at the very faint flux limits ( down to @xmath144 mjy ) probed by the bima experiment other source populations undetectable at brighter fluxes
the detailed analysis of @xcite shows that high - redshift dusty proto - spheroidal galaxies in the main phase of star formation @xcite are contributing to arcmin cmb fluctuations even at frequencies as low as @xmath145 ghz .
moreover , if these sources show the mm excess detected in several galactic clouds as well as in nearby galaxies , their contribution to cmb fluctuations would be higher and , when summed in quadrature to the contribution of radio selected sources , would imply a signal only marginally consistent with the bima data .
as an application , we have combined current wmap data on eps @xcite with our 2d simulation code for creating a _ realistic map _ of eps in the submillimeter domain .
obviously , it is necessary to follow a four step process : first of all , we use the epss-2d code for simulating an all sky map at 70 ghz by using the same assumptions made in 3.2.1 and distributing only sources with fluxes @xmath95 jy .
this flux density roughly corresponds to the source detection limit of the wmap k and q channels , in which the wmap survey reaches the highest number of detections .
subsequently , the flux densities , @xmath146 , of all wmap sources have been converted to the corresponding values at _ planck _ lfi frequencies , by using the spectral indexes published by @xcite . among all fluxes , the ones measured at 61 ghz by the wmap v channel have been used as the reference values for converting them to 70 ghz , given that this is the wmap channel closest to the frequency of the map . in the case of a no detection in the v channel , the flux measured in the q channel has been used ; alternatively , the density flux of the k channel has been exploited , in such a way that we could convert all fluxes to our frequency . then , every source in the wmap 1year catalogue has been placed in that specific pixel of the healpix ( nside=2048 ) pixelization scheme corresponding to the published galactic coordinates of the source .
notice that , after having determined the position of each wmap source , i.e. the corresponding healpix pixel , we have _ put to zero _ the flux previously simulated by our code in that particular pixel .
finally , we assign to each `` zero flux '' pixel the flux ( or , equivalently , the temperature ) of the wmap source corresponding to that position .
moreover , we have also checked that all the simulated fluxes which fall inside a ring of the dimension of the wmap fwhm around each wmap source are not greater than the corresponding error in flux of the wmap 1-year eps catalogue .
this choice guarantees that we can exactly recover the fluxes of all the brightest
wmap detected sources from our map , shown in figure 7 .
this map is a very realistic realization of the eps sky at 70 ghz given that it combines the wmap catalogue ( bright detected sources ) and realistic 2d simulations ( faint sources ) , as discussed before .
moreover , as soon as the 2-year wmap catalogue shall be available , this same map , as well as other maps of eps at other frequencies @xmath147 ghz , can be easily and rapidly updated .
we have presented predictions on the angular power spectrum of eps by 2d simulated maps in which extragalactic sources are distributed with correlated positions in the sky .
the code , named epss-2d , is able to simulate both sky patches and all sky maps taking short cpu process times in a common workstation ( see 2.2.1 ) .
we want to stress again that we adopted a purely _ phenomenological _ approach given that our main purpose was the definition of a fast and flexible tool for simulating 2d maps of eps , under the most general assumptions on source counts and on the angular correlation functions of point sources .
the present outcomes are , obviously , _
model dependent_. on the other hand , they try to take into account all current data on eps counts as well as on source clustering .
all these simulated maps can be very useful for studying the effect of clustered eps , in view of estimating the cmb angular power spectrum and bispectrum .
\1 ) 2d maps are created by first making a poisson density field , corresponding to the differential counts of eps , @xmath23 , which have been determined or estimated at the frequency under study ; subsequently , we modify in the fourier space this `` white noise '' density field according to some angular power spectrum , @xmath0 , the most reliable one for that specific source population , whose differential counts have been used ; as a final step , we distribute fluxes on the density field map , under the condition that @xmath47 fluxes taken from the differential counts fall in a pixel whose number of sources is @xmath47 .
this method , discussed in detail in 2 , proves `` safe '' given that we are always able to recover the input source counts and the input @xmath0 .
moreover , the proposed method allows us to estimate also the contribution of clustered eps to the cmb bispectrum @xcite .
\2 ) the method allows , in principle , the simulation of as many eps populations as needed , provided that their contribution to the differential counts at a given frequency is known . on the other hand , in almost all the cases it will be sufficient to only distribute in the sky those eps coming from the dominant source population at a given frequency , thus making the simulation simpler and faster .
this is generally the case , except in some very particular frequency range , in which number counts at fluxes relevant for a particular experiment are contributed by a comparable number of sources coming from two ( or more ) source populations which show different clustering properties ( see 2.2.2 ) .
\3 ) by using the toffolatti et al . cosmological evolution model for eps and by applying the @xmath148 of @xcite we have estimated the temperature angular power spectrum of clustered eps at wmap and _ planck _ lfi frequencies .
our current results confirm that the extra power due to clustering of eps is _ always small _ in comparison with the poisson term in all cases for which no subtraction of bright sources is applied ( see 4.1 ) .
as extensively discussed in the body of the paper , the present outcome is in full agreement with all the main studies on the subject .
\4 ) on the other hand , we find that at frequencies @xmath149 ghz , the contribution of the clustering term , @xmath150 , to the total confusion noise can be , very probably , the dominant one , thus confirming previous theoretical predictions .
this result is mainly determined by the very steep slope of eps counts at sub
mm wavelengths combined with the strong clustering signal inferred for high redshift spheroidal galaxies and scuba sources @xcite .
\5 ) our current simulations allows us to also estimate the contribution of clustered eps to the excess signal recently detected at arcmin scales by the cbi and bima fields . our current findings , which as stated before
are model dependent , suggest that clustering of eps can give a small but not negligible contribution to this excess .
\6 ) we present here , as an example , a realistic all sky map of clustered eps sources at 70 ghz .
this map has been obtained by combining the wmap point source catalogue , converted to 70 ghz , and a map of fainter eps at the same frequency , simulated by the epss2d code .
as previously discussed , this map and other ones at wmap and _ planck _ lfi frequencies are easily simulated , thanks to the currently available wmap 1year data .
the lack of data at higher cmb frequencies makes a map of this kind much more uncertain .
finally , we remind that we can very easily update our current analysis and results : e.g. , by taking into account more source populations , for providing more accurate estimates on the @xmath54s of eps . as for the simulated all - sky maps
, we can easily update them by the epss2d code and by exploiting the 2-year wmap source catalogue in the next future .
it is important to stress again that current and future applications of these maps are manifold : first of all , they allow the direct estimate of @xmath54s due to eps , under very general assumptions ; they shall be useful in testing the capability of new algorithms for the detection of eps in presence of correlated positions in the sky ; they shall also allow us to test the efficiency of component separation techniques ; eventually , they can be of some help in the study of non gaussian signatures in residue cmb maps . therefore
, we intend to make publicly available the maps created by our epss2d simulation code .
we hope that they shall be useful to all the research community devoted to the development of current as well future cmb anisotropy experiments .
we thank an anonymous referee for useful comments and suggestions which helped us in improving the final presentation of this paper .
we are grateful to j.m .
diego , e. martnez
gonzlez , p. vielva and r.b .
barreiro for very stimulating discussions as well as for some useful remark on an early version of this paper .
we are indebted to c. burigana for a useful suggestion which helped in the final definition of the all sky map at 70 ghz as well as for the .ps image of the same map .
we also thank : g.l .
granato for providing us with his model counts of eps at planck hfi frequencies ; m. negrello for his useful collaboration in checking the agreement between his theoretical predictions and our simulations on the angular power spectrum of eps ; g. de zotti for reading carefully the final version of the paper .
we acknowledge partial financial support from the spanish mcyt under project esp200204141c0301 .
jgn acknowledges a fpu fellowship of the spanish ministry of education ( mec ) . this work has used the software package healpix ( `` hierarchical , equal area and isolatitude pixelization of the sphere '' , http://www.eso.org/science/healpix ) ,
developed by k.m .
gorski , e.f .
hivon , b.d .
wandelt , j. banday , f.k .
hansen and m. bartelmann .
argeso , f. , gonzlez - nuevo , j. , and toffolatti , l. , 2003 , , 598 , 86 barcons , x. , 1992 , , 396 , 460 baugh c. m. , and efstathiou g. , 1993 , , 265 , 145 bennett , c. l. , et al . , 2003 , , 148 , 97 blake , c. , and wall , j. , 2002 , , 329 , l37 blain , a.w . , and longair , m.s . , 1993 , , 264 , 509 colless , m. , et al .
( 2dfgrs team ) , 2001 , , 328 , 1039 daddi , e. , broadhurst , t. , zamorani , g. , cimatti , a. , rottgering , h. , and renzini , a. , 2001 , a&a , 376 , 825 daddi , e. , et al . , 2003 , , 588 , 50 danese , l. , de zotti , g. , franceschini , a. , and toffolatti , l. , 1987 , , 318 , l15 dawson , k.s . , holzapfel , j.e . ,
carlstrom , j.e . , joy , m. , la roque , s.j . ,
miller , a.d . , and nagai , d. , 2002 , , 581 , 86 de zotti g. , franceschini a. , toffolatti l. , mazzei p. , danese l. , 1996 , , 35 , 289 de zotti , g. , toffolatti , l. , argeso , f. , davies , r.d . , mazzotta , p. , partridge , r.b . , smoot , g.f . , and vittorio , n. , 1999 , in _
3k cosmology : ec - tmr conference _ , aip proc .
cp476 , l. maiani , f. melchiorri and n. vittorio eds . , p.204 de zotti , g. , granato , g.l . , silva , l. , maino , d. , and danese , l. , 2000 , , 354 , 467 de zotti g. , perrotta f. , granato g.l .
, silva l. , ricci r. , baccigalupi c. , danese l. , toffolatti l. , 2003 , in `` srt : the impact of large antennas on radioastronomy and space science '' , sif conf .
81 , n. damico , f. fusi pecci , i. porceddu , and g. tofani eds . , italian physical society , vol .
81 , p.57 ( astro - ph/0204038 ) de zotti , g. , burigana , c. , cavaliere , a. , danese l. , granato , g.l . ,
lapi , a. , platania , p. , and silva , l. , 2004 , preprint ( astro - ph/0401191 ) .
dunlop , j.s . , and peacock , j.a . , 1990 , , 247 , 19 dunlop , j.s . , 2001 ,
, 45 , 609 franceschini , a. , toffolatti , l. , danese , l. , and de zotti , g. , 1989 , , 344 , 35 franceschini , a. , toffolatti l. , mazzei p. , danese l. , and de zotti g. , 1991 , a&a suppl .
series , 89 , 285 gawiser , e. , and smoot , g. f. , 1997 , , 480 l1 giavalisco , m. , steidel , c.c . ,
adelberger , k.l . ,
dickinson , m.e . ,
pettini , m. , kellogg , m. , 1998 , , 503 , 543 grski , k.m . ,
banday , a. , hivon , e. , wandelt , b.d .
, 2002 , astronomical data analysis software and systems xi , asp conference proceedings , vol .
281 , d.a .
bohlender , d.durand , and t.h .
handley eds .
granato , g. l. , silva , l. , monaco , p. , panuzzo , p. , salucci , p. , de zotti , g. , and danese , l. , 2001 , , 324 , 757 granato , g. l. , de zotti , g. , silva , l. , bressan a. , and danese , l. , 2004 , , 600 , 580 guiderdoni , b. , hivon , e. , bouchet , f. , and maffei , b. , 1998 , , 295 , 877 halverson , n.w . , et al . , 2002 , , 568 , 38 herranz , d. , kuruolu , e.e . , and toffolatti , l. , 2004 , , 424 , 1081 hinshaw , g. , et al . , 2003 , , 148 , 135 holland , w.s . , et al . , 1999 , , 303 , 659 ivison , r.j . , et al . , 2002 , , 337 , 1 knox , l. , cooray , a. , eisenstein , d. , and haiman , z. , 2001 , , 550 , 7 komatsu , e. , et al . , 2003 , , 148 , 119 kuo , c.l . , ade , p.a.r
bock , j.j .
, et al . , 2004 , , 600 , 32 loan , a. j. , wall , j. v. , and lahav , o. , 1997 , , 286 , 994 magdwick , d.s . , et al . , 2003 , , 344 , 847 magliocchetti , m. , maddox , s.j . , lahav , o. , and wall , j. , 1998 , , 300 , 257 magliocchetti , m. , moscardini , l. , panuzzo , p. , granato , g.l . , de zotti , g. , and danese , l. , 2001 , , 325 , 1553 magliocchetti , m. , et al . , 2004 , , in press ( astro - ph/0312160 ) mandolesi , n. , et al . , 1998 , proposal submitted to esa for the `` planck '' low frequency instrument ( lfi ) martnez v.j
. , saar e. , 2001 , statistics of the galaxy distribution , chapman & hall , crc , london , uk .
mason , b.s .
, et al . , 2003 , , 591 , 540 matarrese , s. , coles , p. , lucchin , f. , and moscardini , l. , 1997 , , 286 , 115 mather , j.c .
, fixsen , d.j . ,
shafer , r.a .
, mosier c. , and wilkinson , d.t . , 1999 , , 512 , 511 moscardini , l. , coles , p. , lucchin , f. , and matarrese , s. , 1998 , , 299 , 95 negrello , m. , magliocchetti , m. , moscardini , l. , de zotti , g. , granato , g.l . , and silva , l. , 2004 , , 352 , 493 negrello , m. , magliocchetti , m. , moscardini , l. , de zotti , g. , and danese , l. , 2004 , , submitted ( astro - ph/0406388 ) park , c. g. , park , c. , and ratra , b. , 2002 , , 568 , 9 peacock j.a .
, 1997 , cosmological physics , cambridge university press , cambridge , uk .
peacock , j.a .
, et al . , 2000 , , 318 , 535 peebles , p.j.e . ,
1993 , principles of physical cosmology , princeton university press , princeton ( u.s.a . ) perrotta , f. , magliocchetti , m. , baccigalupi , c. , bartelmann , m. , de zotti , g. , granato , g.l . , silva , l. , and danese , l. , 2003 , , 338 , 623 pierpaoli , e. , and perna , r. , 2004 , mnras , in press ( astro - ph/0407561 ) puget , j .- l . , et al . , 1998 ,
proposal submitted to esa for the `` planck '' high frequency instrument ( hfi ) readhead , a.c.s . , mason , b.s .
, contaldi , c.r . ,
et al . , 2004 , , 609 , 498 rowan
robinson , m. , and fabian , a.c .
, 1974 , , 167 , 419 scott , d. , and white , m. , 1999 , a&a , 346 , 1 scott , d. , et al . , 2000 , a&a , 357 , l5 smail , i. , ivison , r.j . ,
owen , f.n . ,
blain , a.w . , and kneib , j.p .
, 2000 , , 528 , 612 sokasian , a. , gawiser , e. , and smoot , g. f. , 2001 , , 562 , 88 sunyaev , r.a . , and zeldovich , ya.b . , 1980 , , 18 , 537 taylor , a. c. , grainge , k. , jones , m. e. , pooley , g. g. , saunders , r. d. e. , and waldram , e. m. , 2001 , , 327 , l1 taylor , a. c. , et al . , 2003 , , 341 , 1066 tegmark , m. , and efstathiou , g. , 1996 , , 281 , 1297 tegmark , m. , de oliveira costa , a. , 1998 , , 500 , l83 tegmark , m. , et al . , 2002 , , 571 , 191 toffolatti , l. , et al . , 1995 , , 32 , 125 toffolatti , l. , argeso gmez , f. , de zotti , g. , mazzei , p. , franceschini , a. , danese , l. and burigana , c. , 1998 , , 297 , 117 toffolatti , l. , argeso , f. , de zotti , g. , and burigana , c. , 1999 , in asp conf
series , vol .
181 , _ microwave foregrounds _ ,
a. de oliveira - costa and m. tegmark eds .
, p.153 toffolatti , l. , gonzlez - nuevo , j. , negrello , m. , de zotti , g. , silva , l. , granato , g.l . , and argeso , f. , 2004 , , submitted , astro - ph/0410605 tucci , m. , martnez - gonzlez , e. , toffolatti , l. , gonzlez
nuevo , j. , and de zotti , g. , 2004 , , 349 , 1167 vielva , p. , martnez - gonzlez , e. , cayn , l. , diego , j.m . ,
sanz , j.l . , and toffolatti , l. , 2001 , , 326 , 181 vielva , p. , martnez - gonzlez e. , gallegos , j. e. , toffolatti , l. , and sanz , j.l . ,
2003 , , 344 , 89 waldram , e. , pooley , g.g . ,
grainge , k.j.b . ,
jones , m.e . , saunders , r.d.e . , scott , p.f . , and taylor , a. , 2003 , , 342 , 915 webb , t.m . , eales , s.a . , lilly , s.j . ,
clements , d.l . , dunne , l. , gear , w.k . ,
ivison , r.j . , flores , h. , and yun , m. , 2003 , , 587 , 41 | we present predictions on the angular power spectrum of cosmic microwave background ( cmb ) fluctuations due to extragalactic point sources ( eps ) by using a method for simulating realistic 2d distributions of clustered eps .
both radio and far ir selected source populations are taken into account . to analyze different clustering scenarios , we exploit angular power spectra of eps , @xmath0 , estimated either by data coming from currently available surveys or by means of theoretical predictions . by adopting the source number counts predicted by the toffolatti et al .
( 1998 ) evolution model
capable of accounting well for the available data at radio cm wavelengths we are able to reproduce current data on the two point angular correlation functions , @xmath1 , of radio sources
. we can confirm that the detection of primordial cmb anisotropies is not hampered by undetected clustered sources at frequencies @xmath2 ghz . on the other hand ,
our current findings show that at higher frequencies the clustering signal could severely reduce the detectability of intrinsic cmb anisotropies , thus confirming previous theoretical predictions .
we also show that unsubtracted eps can account for the excess signal at high multipoles detected by recent cmb anisotropy experiments .
moreover , the additional power due to the clustering of sources gives rise to a small but not negligible contribution to the same excess signal . as a final result
, we also present an example of a currently feasible _ realistic map _ of eps at 70 ghz , by taking into account data on bright detected sources as well as the previously quoted model for number counts .
eventually , these simulated sky maps can prove very useful to test the efficiency of component separation techniques , the capability of new algorithms for the detection of eps and the appearance of non gaussian signatures in `` residue '' cmb maps , in the presence of sources which are not poisson distributed in the sky . -0.8truecm |
we begin by briefly reminding that a conjugated continuous effort has been made to relate the emergence of the collective energy dissipation through one and two - body nuclear processes with the chaotical behaviour of nuclear systems . a few options in choosing the collective oscillation frequencies have come into focus in the past years , in connection with the onset of chaoticity for `` nuclear billiards '' .
first of all , the issue of dissipation into thermal motion of the adiabatic collective vibrational energy of the potential well was treated for several multipolarities by burgio , baldo _ et al . _ . on the other hand , when trying to associate different vibration frequencies to various nuclear processes , the path to chaos was found to be changed with the order of multipole @xcite .
this paper was intended to bring a quantitative argumentation , based on a systematic study of the configuration and kinetic energy spaces , power spectra , informational entropies , and largest lyapunov exponents .
the study was done in completion of a few qualitative types of analysis previously presented @xcite : sensitive dependence on the initial conditions , single - particle phase space maps , fractal dimensions of poincare maps , and autocorrelation functions .
in short , we remind that , by studying the nucleonic dynamics in a woods - saxon potential , one can find an increase of the chaotical degree of the system behaviour as raising the frequency of 2d wall oscillation . the main result of @xcite was reported in relation with an intermittent route to chaos for the monopole vibrations close to the resonance phase of a nuclear interaction .
still , we mention that the purpose of these two coupled etudes was only to emphasize the detection of such intermission for the `` nuclear billiards '' and not to establish its type according to @xcite , nor to compare it with other intermittency patterns from known experimental results .
we continue the study on a classical dynamical system proposed by burgio , baldo _ et al .
_ , system composed of a number of @xmath0 nucleons with no charge , spin , or internal structure . a two - dimensional deep woods - saxon potential well , regarded as `` nuclear billiard '' , is periodically hit by the nucleons .
the bohr hamiltonian in polar coordinates is a sum of two components : kinetic ( @xmath1 ) and potential ( @xmath2 ) , the kinetic one decoupling into radial ( @xmath3 ) , centrifugal ( @xmath4 ) , and collective terms ( @xmath5 ) : @xmath6 @xmath7 the phase space is defined by particle and collective momenta and their conjugate coordinates : @xmath8 , @xmath9 and @xmath10 .
the collective coordinate @xmath11 oscillates with @xmath12 frequency , the inglis mass @xmath13 is equal to @xmath14 , and the nucleon mass : @xmath15 . for the time being we are not interested in studying the nucleon dynamics beyond the woods - saxon barrier : @xmath16 } ,
\ ] ] and therefore we choose a deep well : @xmath17 and accordingly , a low value for the diffusivity coefficient : @xmath18 . when considering the two - dimensional case ,
the frontier of the collective motion is described as a function of the collective variable @xmath11 and of the legendre polynomials
@xmath19 @xcite : @xmath20.\ ] ] the oscillation degree of the potential well @xmath21 is considered for the monopole @xmath22 , dipole @xmath23 and quadrupole case @xmath24 . if the surface has a stationary behaviour , or whenever one takes into account the uncoupled hamilton equations ( uce ) for the particle : @xmath25 and collective degrees of freedom ( _ d.o.f .
_ ) : @xmath26 @xmath27 has a fix value , chosen for consistency with previous papers @xcite as @xmath28 .
a runge - kutta type algorithm ( order 2 - 3 ) with an optimized step size was used for solving the system of differential equations , while keeping the absolute errors for the phase space variables under @xmath29 and conserving the total energy with relative error : @xmath30 ( fig .
[ fig:1 ] ) .
we imposed the equilibrium condition between the pressure exerted by the particles and the mechanical pressure of the wall @xcite and thus obtained for the initial equilibrium value of the collective variable , perturbed with a small value : @xmath31 where @xmath32 @xcite is the two - dimensional kinetic energy and also the temperature of the nuclear system , when considering the natural system of units ( @xmath33 ) .
we carry on the study begun in @xcite by gradually changing the degree of vibration of the potential wall from a slow motion ( adiabatic state ) to a rapid one ( the so - called resonance state of the interaction ) . in the first case ,
the collective motion is described by a radian frequency smaller than @xmath34 .
the latter is dominated by frequencies close to the one - nucleon collisional frequency : @xmath35 the physical motivation for studying the one - nucleon chaotical dynamics in the `` nuclear billiard '' , ranging the frequencies from the adiabatic to the resonance regime of a nuclear interaction , is explained in some detail in @xcite .
we briefly remind that the process of nuclear multifragmentation can be viewed as a resonance process and that for smaller excitation energies , nuclear evaporation or breakup of a projectile nucleus occurs when the energy is shared between the collective and one - nucleon degrees of freedom . by using a few types of analyses : sensitive dependence on the initial conditions , single - particle phase space maps , fractal dimensions of poincare maps and autocorrelation functions
, we emphasized that an intermittent route to chaos is observed in the monopole case when increasing the vibrational frequency to @xmath36 @xcite . in the resonance phase of the interaction the onset of chaotical behaviour
was found to be earlier than at any other adiabatic oscillations of the woods - saxon potential well .
we present here other methods @xcite promoting the idea that the degree of chaoticity increases when moving from the adiabatic to the resonance regime : analyses of the configuration and kinetic energy spaces , power spectra , generalized informational entropies and lyapunov exponents .
furthermore , we try to identify possible pathways to chaos , including the intermittent one , previously put in evidence @xcite . in order to establish
if a specific physical system presents a chaotic dynamics and to identify possible routes to chaos we analyzed the behaviour of a small bunch of trajectories in the configuration and in the kinetic energy space , respectively .
for example , we took five trajectories separated by an @xmath37 aperture , while keeping the rest of the initial phase space variables constant and let the system evolve over a given time @xmath38 .
for the transient stages from adiabatic to resonance , the temporal evolution of an initially confined trajectory bundle was studied for the monopole and dipole oscillation modes of the potential wall and also for the limit situation , in which the individual and collective degrees of freedom remain uncoupled ( figs .
[ fig:2 ] and [ fig:3 ] ) .
the configuration space revealed a high degree of symmetry in @xmath39 plane in both cases , @xmath40 uncoupled nonlinear differential equations and monopole ( left and middle panels ) .
also , that the central zone remained uncovered , reflecting the conservation of the nucleon angular momentum .
another important conclusion was issued from the definition of the stability concept of a dynamical system . for the aforementioned cases the dispersed trajectory pack periodically regroups on the frontier that delimits the forbidden zone of the phase space .
this type of behaviour corresponds with the definition of stability given by poisson ( for e.g. , in @xcite ) .
we will herewith remind that the poisson stability defines as steady the movement of a particle system of which configuration comes close , from time to time , to the initial position . at a first glance , on the simple uce case one can distinguish two extremities of the radius of the particle periodic motion in the 2d potential well : @xmath41 , and @xmath42 .
these values correspond to the roots of the boundary equation : @xmath43 for a given one - nucleon energy @xmath44 , when the radial component of velocity vanishes ( @xmath45 ) .
the analysis of the particle motion in the configuration space is similar to that applied to any system with bound unclosed trajectories .
the nucleonic motion takes place within a circular crown ( the so - called _ annulus _ ) determined by the concentric circles of @xmath46 and @xmath47 radii .
the trajectory is symmetric about any turning point .
for @xmath48 we descry in the uce case as much as @xmath49 distinct apocenters ( at @xmath50 ) .
these are correlated with a number of @xmath51 complete and one incomplete revolutions about the center of the force field ( _ i.e. _ @xmath49 straight lines before @xmath52 changes its sense of variation ) .
the particle completely sweeps over twice the 2d configuration space after @xmath53 .
however , we notice that the bound trajectories are open , which means that the orbits never pass twice through a given point ( see fig .
[ fig:2 ] ) , which is in concordance with bertrand s theorem .
we briefly remind that for a bound orbit to be closed , the angle between two consecutive apocenters must be : @xmath54 _ i.e. _ after @xmath55 revolutions about the center , the radius vector should sweep out a multiple @xmath56 of @xmath57 radians . for the uce case above considered
we consequently obtain : @xmath58 radians ( fig .
[ fig:2 ] - left column ) .
the kinetic energy points are displayed in right isosceles triangular shaped patterns ( fig .
[ fig:3 ] ) , whose hypotenuses are described by eqs . ( 11 ) and ( 12 ) , for the two specific non - chaotical situations : uce and the intermittent monopolar `` window '' emerged at @xmath36 : @xmath59 @xmath60 moreover , for the intermittency frequency of monopole oscillations , one can notice a second smaller segment with the same negative slope : @xmath61 for the uncoupled differential equations there are a couple of extreme values for the centrifugal kinetic energy : @xmath62 , and @xmath63 , associated with @xmath64 and respectively , with @xmath65 ( left column of fig .
[ fig:3 ] and eq .
( 1 ) ) . as for the monopolar intermittency ,
we can distinguish just five distinct values for the @xmath66 : @xmath67 , @xmath68 , @xmath69 , @xmath70 , and @xmath71 ( central plot of fig . [ fig:3 ] ) , correlated with stationary radii : @xmath72 , @xmath73 , @xmath74 , @xmath75 , and @xmath76 ( eq . ( 1 ) , fig
. 2 of @xcite , and central plot of fig .
[ fig:2 ] ) .
thus , the nucleonic motion for the intermittent case is composed of alternated revolutions about the force field centre , forming a cyclic symmetrical structure , for e.g. , @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath80 , @xmath79 , @xmath78 , @xmath77 , and so on ( fig . [ fig:2 ] ) .
this behaviour can be easily verified through the sensitivity dependence on the initial conditions analysis , previously presented ( third column of fig .
2 - @xcite ) .
it should also be mentioned that , following this radius alternation , the nucleon covers in 2d configuration space @xmath82 radians after @xmath83 full revolutions in almost @xmath84 .
concluding , we highlight once more , that ordered , non - chaotical events , exhibit periodical symmetrical patterns in the configuration and kinetic energy spaces .
this was shown to be a characteristic feature of the uncoupled nonlinear hamilton equations case and also , of the steady , intermittent behaviour arisen in the monopole case at @xmath85 vibrational radian frequency . a tendency to compactly fill the kinetic energy space when increasing the monopolar vibrations ( from @xmath34 to @xmath86 )
was observed , except for the intermittency situation above described .
for the @xmath87 oscillation mode of the potential well , it seems that at the same frequency ( @xmath36 ) a somewhat intermittent behaviour could also come out , but this was proved to be elusive , as verified when reverting to this issue with the help of informational entropies and lyapunov exponents and analyzing the system on longer time periods . in order to better distinguish between a multiple periodical behaviour that can also exhibit an erratic pattern and chaos we used the fourier transform of the analyzed signals : @xmath88 @xmath89 for a multiple periodical movement the power spectrum : @xmath90
will only contain a number of discrete lines : the fundamental frequencies of the system and their associated sets of harmonics , while the chaotical behaviour is completely aperiodical and is represented by a continuous or quasi - continuous broadband .
the obtained results are presented in figures [ fig:4]-[fig:7 ] . as a persistent feature of the physical system analyzed one should mention that for the monopole and dipole deformation degrees of the potential well ( figures [ fig:5 ] and [ fig:6 ] ) the chaotic behaviour increases in time , thus confirming previous results .
the transition towards a chaotic regime was put again in evidence once passing from the adiabatic to the resonance stage of the interaction .
the power spectra reveal , as expected , the intermittent feature of the transition in the monopolar case at @xmath91 .
this can be detected for periods of time large enough ( @xmath92 ) to positively identify chaotic patterns , by transition from a quasi - continuous spectrum of the one - nucleon radial coordinate ( @xmath93 ) to a discrete periodical one , containing fundamental frequencies of the system and its harmonics ( @xmath91 ) and again to a continuous spectrum at the resonance vibrational frequency ( @xmath94 ) ( fig .
[ fig:5 ] - right panels ) .
the order - chaos - order - chaos sequence can be also spotted out for the monopole oscillations in the power spectra of the collective degree of freedom ( fig .
[ fig:7 ] - second column ) . the temporal series of the radius variable show a symmetrical sawtooth waveform for the uncoupled situation at any chosen vibration frequency , and also for the monopole case at adiabatic collective oscillations . for the rest , in general an asymmetrical sawtooth form defines the series , but sometimes , more complicated patterns appear at higher multipole orders ( figures 1 - 4 - @xcite ) .
the difference between two successive maxima in the temporal series of the radius variable for the uce case is : @xmath95 ( left column of fig .
[ fig:2 ] ) and can be also obtained from the sensitive dependence on the initial conditions analysis ( see fig . 1 - @xcite ) . the fundamental frequency for the radial sawtooth temporal series : @xmath96 and its first three harmonics : @xmath97 , @xmath98 , and @xmath99 ,
can be easily traced down in figure [ fig:4 ] .
as for the `` window '' of intermittency at @xmath100 , we obtained : @xmath101 ( fig . 2 - @xcite and central plot of current fig .
[ fig:2 ] ) .
this gives the corresponding fundamental frequency : @xmath102 and its associated harmonics : @xmath103 , @xmath104 , and @xmath105 ( fig .
[ fig:5 ] ) . in order to further investigate route to chaos ,
we paid attention to the time evolution of the generalized informational entropy ( or shannon entropy ) , introduced as usually : @xmath106 @xmath107 being the number of gradually occupied cells until the time @xmath108 .
this type of entropy is actually a number which quantifies the time rate of information production for a chaotic trajectory @xcite .
we consider in the first place the case of a particle that at every moment occupies a cell of the two - dimensional lattice phase space with a @xmath109 probability : @xmath110 where : @xmath111 @xmath112 , @xmath113 , and @xmath114 are the number of bins of the @xmath115 lattice . for @xmath116 is a constant of motion for the monopole and the uce cases , we use for comparisons only these three phase space variables . as an alternative measure for the above defined entropy we also used the cumulative filling percentage of the one - nucleon phase space : @xmath117 in the first place , for a given wall frequency of vibration and for a certain multipolarity ( here , for @xmath94 and @xmath100 ) , we studied the dependence of the shannon entropy with the number of bins .
a clear tendency for smoothing the entropy curve was found when decreasing the bin .
a reduced number of cells ( @xmath118 ) is characterized by an entropy formed from a small number of high - amplitude heaviside functions .
as the number of bins increases ( for e.g. , here to @xmath119 ) , the entropy gets a more realistic representation , being composed of a superior number of low - amplitude step functions ( fig .
[ fig:8 ] ) .
moreover , the filling percentage @xmath120 of the one - nucleon phase space maps can drastically differ with the size of the bin .
thus , after the system evolved over @xmath121 , a phase space with @xmath122 bins is entirely covered , @xmath123 bins can be filled in with @xmath124 probability , a @xmath125 filling percentage for @xmath126 cells can be found , and we counted only as much as @xmath127 bins occupied out of a total of @xmath128 ( _ i.e. _ @xmath129 ) . at a first glance one can identify a series of entropy plateaus , which could be put in correspondence with stationary or quasi - stationary thermodynamic values of the system if a large number of particles would be under study .
some of them will vanish when considering a large number of bins .
however , those surviving for @xmath130 could be associated with stationary nucleonic states in the chosen potential well in the limit of a large number of degrees of freedom . for a given 2d phase space lattice formed of @xmath131 bins we present in figures [ fig:9]-[fig:12 ] a comparison between the informational entropies of the physical system in study , starting from the adiabatic stage of interaction and gradually increasing the vibrational wall frequency towards the dipole resonance value , @xmath94 .
the slopes for the resonance frequency case were found to be significantly higher than for the adiabatic one ( @xmath93 ) for all multipolarities involved .
another comparison revealed significant differences between the onset times of the quasi - constant shannon entropy values for all cases taken into consideration .
thus , for four vibrational radian frequencies and for four coupling modes of the hamilton equations we show the informational entropy values after @xmath132 ( table [ tab:1 ] ) and the associated phase space filling degrees ( table [ tab:2 ] ) . also , in table [ tab:3 ] , are presented the periods of time after which the filling percentages @xmath120 equal unity .
.[tab:1]the computed @xmath133 of the @xmath134 one - particle phase space maps at several multipolarities and frequencies of wall vibration [ cols="^,^,^,^,^ " , ] we begin the analysis with the uce case .
the single and collective uncoupled _ d.o.f .
_ give birth to a quasi - laminar behaviour with a weak development of chaotic states .
the one - particle informational entropy shows an identical evolution , no matter the frequency chosen .
the orbit covers , after @xmath135 , only @xmath136 of the entire lattice ( table [ tab:2 ] and figure [ fig:9 ] ) and does not reach @xmath137 , even after @xmath138 ( table [ tab:3 ] ) . also , the phase space is not covered up by all five trajectories for the whole range of @xmath139 considered , when analyzing @xmath140 ( fig .
[ fig:13 ] and table [ tab:4 ] ) . for the dipole oscillations mode , at @xmath141
, it appears that , after only @xmath135 , the entropy closes in upon its maximum value : @xmath142 ( fig . [ fig:11 ] and table [ tab:1 ] ) .
however , on long periods of time , the real tendency is towards filling up the nucleonic phase space as rapid as the vibrational frequency is increased ( table [ tab:3 ] ) .
the exact pattern is repeated when studying the shannon entropy for closeby nucleonic trajectories ( fig .
[ fig:15 ] and table [ tab:4 ] ) .
we found quite the same feature for the monopole case , with exception for the intermittent `` window '' at @xmath91 ( tables [ tab:1 ] , [ tab:2 ] and fig .
[ fig:10 ] ) .
the occupying rate is so small in the intermittent zone , that just at @xmath143 , the particle would have covered the whole phase space ( see table [ tab:3 ] ) .
a similar conclusion can be drawn from table [ tab:4 ] and fig .
[ fig:14 ] ( with a double temporal scale scanned for the intermittent frequency ) .
the trajectory pack informational entropy reaches its highest value after the longest one - particle evolution time of all : @xmath144 .
the quadrupole oscillation also reveals an apparent intermittent pattern , this time at @xmath145 .
we call it intermittent because after @xmath135 the nucleon fills in only @xmath146 of the total number of bins ( figure [ fig:12 ] and table [ tab:2 ] ) , and a longer time than @xmath147 is required to get to @xmath148 ( table [ tab:3 ] )
. however , this behaviour can be a misleading one , the shannon entropy for a trajectory bunch showing exactly the opposite ( see fig . [ fig:16 ] and table [ tab:4 ] ) , after @xmath149 the orbits being completely dispersed .
we furthermore presented another quantitative analysis : the temporal evolution of the lyapunov exponents , @xmath150 .
as previously shown , initial adjacent points in the phase space @xmath151 , can generate in time separated trajectories @xmath152 .
when studying the evolution of a single phase space parameter , the one - dimensional lyapunov exponent takes the form : @xmath153 the generalization for obtaining the multi - dimensional lyapunov exponent is then straightforward : @xmath154^{2}\right)^{\frac{1}{2}}}{0.01},\ ] ] where the sum is taken over all @xmath155 squared differences between final @xmath156 and initial @xmath157 one - nucleon phase space variables .
integration times of the order of @xmath158 fm / c exclude errors when computing the lyapunov exponents .
in short , we here remind that the trajectories can be classified as function of the lyapunov exponents .
thus , one can distinguish periodical behaviours , for @xmath159 , dissipative movements with a fixed point or a basin of attraction @xmath160 , and aperiodical chaotic states @xmath161 , when the iterative discrete evolution of the solution series ( eqs .
5 and 6 ) leads to a chaotic pattern .
another way of measuring the system sensitivity to initial conditions is to compute the largest lyapunov exponent ( lle ) .
usually a couple of methods can be employed , one based on the time dependence of the multi - dimensional lyapunov exponent , the other on wolf s standard method that uses a gram - schmidt reorthonormalization of the tangent vectors @xcite . in the latter , the lle is obtained by taking the asymptotic value of the multi - dimensional lyapunov exponent : @xmath162 still , this method has the disadvantage that the integration times have to be at least an order of magnitude larger than those here considered .
other methods are slightly less efficacious , being more cpu time - consuming when simulating strong chaotic systems @xcite .
we consequently used the first method and noticed the saturation behaviour , _
i.e. _ the arising of a plateau after a certain time @xmath163 ( fig .
[ fig:17 ] ) .
the straight lines represent fits whose slopes match the lle ( table [ tab:5 ] ) .
they are in inverse proportion with the onset times of chaoticity @xmath164 , being a measure of the trajectory decoupling at a microscopic level .
cccc oscillation frequency & @xmath165 & @xmath166 & @xmath167 + @xmath168 & @xmath169 & @xmath170 & @xmath171 + @xmath172 & @xmath173 & @xmath174 & @xmath175 + @xmath176 & @xmath177 & @xmath178 & @xmath179 + @xmath180 & @xmath181 & @xmath182 & @xmath183 + when the single - particle and collective _ d.o.f .
_ remain uncoupled , the lyapunov exponents basically oscillate between two quasi - stationary regimes .
this happens for all vibrational frequencies involved , reflecting a periodical regrouping of orbits in two basins of attraction .
the phase space not being covered , even after a hundred of thousand of @xmath184 , computing the lle becomes futile for this case .
one can remark for dipole oscillations ( fig .
[ fig:17 ] - middle column ) a faster evolution towards reaching saturation states of the 4-dimensional lypaunov exponents , once passing from the adiabatic @xmath185 to the resonance phase of the interaction @xmath186 . in the monopolar case
the intermittency can be easily traced at @xmath85 vibrational frequency ( fig .
[ fig:17 ] - left panels ) . during the intermittent stage ,
independent nearby orbits microscopically diverge with the slowest rate of all : @xmath187 ( table [ tab:5 ] ) . in order to catch the heaving in sight of the stationary plateau at @xmath188 ,
the temporal scale was scanned over @xmath189 .
the study of the quadrupole collective oscillation case confirms the results obtained with all previous analyses .
namely , the neighbouring trajectories deviate one from each other after just @xmath190 at an adiabatic frequency : @xmath34 .
also , when increasing @xmath191 , the lle evolution pattern exactly matches that found with informational entropy measured for a group of orbits ( tables [ tab:4 ] and [ tab:5 ] ) .
we investigated the chaotic nucleonic behaviour in a two - dimensional deep woods - saxon potential well for specific phases of the nuclear interaction . by comparing the order - to - chaos transition for these cases of interest , from adiabatic to resonance regime ,
it was shown that the couplings between the one - particle dynamics and high multipole vibrational modes significantly decrease the onset of the chaotic nucleonic motion towards realistic nuclear interaction time scale .
the quantitative study enfolded a plethora of analyses , pointing out that the paths to chaos for the `` nuclear billiard '' are dissimilar for the studied multipolarities .
for the first two multipole degrees we noticed a more rapid emergence of chaotic states as moving on towards higher radian frequencies of oscillation . when analyzing the system with quadrupole collective deformations of the potential well , an order - strong chaos - weak chaos - order sequence is revealed . still ,
as emphasized in the `` shannon entropies '' subsection , the quadrupole case represents an intricate one , and further analysis would be required before concluding it .
every type of quantitative analysis strengthened previous results regarding the monopolar intermittency route to chaos for the `` nuclear billiard '' .
the collective oscillation frequency for the intermittent behaviour was located prior to the resonance state of interaction ( at @xmath192 ) .
further studies along the above issues are currently in progress .
the used formalism can be improved by adding spin and charge to the nucleons . a semi - quantal treatment of this problem , including pauli blocking effect ,
is hoped to shed more light on the discussed issue in the near future .
d. felea , c. beliu , r.i .
nanciu , al .
jipa , i.s .
zgur , r. mrginean , m. haiduc , a. ghea , and m. ghea , in _ proceedings of the @xmath194 international conference `` nucleus - nucleus collisions '' , strasbourg , 2000 _ , edited by w. norenberg _
( north - holland , amsterdam , the netherlands , 2001 ) p. 222
d. felea , i.v .
grossu , c.c .
bordeianu , c. beliu , al .
jipa , a.a .
radu , c.m .
mitu , and e. stan , `` intermittency route to chaos for the nuclear billiard - a qualitative study '' , phys .
c ( submitted ) .
a. bialas , in _ proceedings of the nato - asi international summer school `` particle production spanning mev and tev energies '' , nijmegen , 1999 _ , edited by w. kittel , p.j .
mulders , and o. scholten , nato science series c : mathematical and physical sciences - vol .
554 ( kluwer academic publishers , nijmengen , the netherlands , 1999 ) . | we extended a previous qualitative study of the intermittent behaviour of a chaotical nucleonic system , by adding a few quantitative analyses : of the configuration and kinetic energy spaces , power spectra , shannon entropies , and lyapunov exponents .
the system is regarded as a classical `` nuclear billiard '' with an oscillating surface of a 2d woods - saxon potential well . for the monopole and dipole vibrational modes we bring new arguments in favour of the idea that the degree of chaoticity increases when shifting the oscillation frequency from the adiabatic to the resonance stage of the interaction .
the order - chaos - order - chaos sequence is also thoroughly investigated and we find that , for the monopole deformation case , an intermittency pattern is again found .
moreover , coupling between one - nucleon and collective degrees of freedom is proved to be essential in obtaining chaotic states . |
the 2df galaxy redshift survey ( 2dfgrs ; maddox , 1998 ) is a large - scale survey aimed at obtaining spectra for 250,000 galaxies to an extinction - corrected limit for completeness of @xmath1 over an area of 2151 square degrees .
redshifts for all the sources brighter than @xmath1 are determined in two independent ways , via both cross - correlation of the spectra with specified absorption - line templates and by emission - line fitting .
the success rate in redshift acquisition is estimated about 95 per cent .
the median redshift of the galaxies is 0.11 and the great majority have @xmath2 .
radio observations come from the first ( faint images of the radio sky at twenty centimetres ) survey ( becker et al .
1995 ) , estimated to be 95 per cent complete at 2 mjy and 80 per cent complete at 1 mjy .
optical counterparts for a sub - sample of first sources have then been obtained by matching together objects included in the radio catalogue with objects coming from the apm survey in the overlapping region @xmath3 and @xmath4 . out of approximately 24,000 radio sources with @xmath0 mjy in the considered area ,
magliocchetti & maddox ( 2001 ) find 4075 identifications in the apm catalogue for @xmath5 ( the limiting magnitude of the apm survey ) and 971 for @xmath6 ( the completeness limit of the 2dfgrs ) , both obtained for a matching radius of 2 arcsecs .
2df data then provided optical spectra for 557 objects with @xmath7 and @xmath0 mjy , corresponding to 53 per cent of the original sample .
this apparent incompleteness is merely due to incomplete sky coverage of the spectroscopic survey .
neither radio nor magnitude biases have been found in the determination of the optical and spectroscopic counterparts of first radio sources .
classes for the optical counterparts of radio sources have been assigned on the basis of their 2df spectra which allowed galaxies to be divided into 4 broad categories . as fig.1 shows , different classes occupy different regions on the @xmath8 and @xmath9-z planes , reflecting the different intrinsic characteristics of the populations which we summarize here : + 1 ) * early - type galaxies * show spectra dominated by continuum which is much stronger than the intensity of any emission line .
this class comprises 289 objects and makes up 52 per cent of the whole spectroscopic sample .
sources belonging to this population tend to have relatively high values for the radio - to - optical ratio , @xmath10 ( @xmath11 ) and are preferentially found for redshifts @xmath12 .
the majority of these sources show rather red colours ( @xmath13 ) , with a tendency to be redder at larger look - back times .
their radio fluxes lie in the range @xmath14 , even though objects are found up to @xmath15 mjy , and optically they appear as relatively faint ( about 60 per cent of the sources has @xmath16 ) .
the @xmath17 relation in this case ( represented by the dashed line in the bottom - left panel of fig .
1 ) is well described by the expression @xmath18 , ( @xmath19 is the luminosity distance ) , with @xmath20 for an @xmath21 , @xmath22 universe , showing that passive radio galaxies are reliable standard candles .
+ 2 ) * e+agn - type galaxies * show spectra typical of early - types plus the presence of ( narrow ) emission lines which indicate the presence of an active nucleus .
there are 61 objects in this class , corresponding to 11 per cent of the spectroscopic sample .
these sources are directly connected to the class of early - type galaxies , even though they show characteristics that are intermediate between pure agn - fuelled sources and star - forming galaxies .
optically they appear as rather faint and closely follow the standard - candle relationship found for early - type galaxies .
however their radio - to - optical ratios are in general as low as those obtained for late - type galaxies .
their radio fluxes are in general quite low and their @xmath23 colours are uniformly distributed between about 0 and 2 .
+ 3)-4 ) * late - type galaxies * and * starbursts * have spectra which show strong emission lines characteristic of star - formation activity , either together with a detectable ( late - type ) or missing ( starbursts ) continuum .
this class comprises 177 objects ( including 10 starbursts ) , which is @xmath24 per cent of the spectroscopic sample .
the radio - to - optical ratios @xmath25 for these sources have values in general smaller than those found for early - type galaxies .
their radio fluxes are rarely brighter than @xmath26 mjy and optically they have bright apparent magnitudes ( very few objects are found with @xmath27 ) .
these sources have quite blue colours , @xmath28 , and are mostly local the majority of them being located within @xmath29 . furthermore , they show a weaker correlation between @xmath9 magnitudes and redshift than found for the previous classes of objects , and do not follow the @xmath17 relation found for early - type galaxies ( illustrated by the dashed line in the bottom panels of fig .
1 ) . + note that we also found 18 seyfert 2 and 4 seyfert 1 galaxies .
for further discussion and the whole list of objects see magliocchetti et al .
( 2001 ) .
some information on the above classes of sources can also be derived from the few optical images showing resolved structures .
for instance , it is interesting to note that the majority of the interacting / merging systems seems to be associated with early - type spectra , typical of pure agn - fuelled sources , suggesting that merging , under appropriate conditions , can trigger agn activity even at low redshifts .
also , as expected , irregulars and spirals preferentially show spectra typical of late - type galaxies ; signatures of interaction and/or merging are seen for members of this latter population as well as for e+agn galaxies .
finally , radio images show that there is a clear trend for extended / sub - structured sources to be associated with absorption systems ( i.e. early - type galaxies ) .
radio powers have subsequently been calculated according to the relation @xmath30 , with @xmath31 angular diameter distance and @xmath32 spectral index of the radio emission , and the local radio lf for objects in the spectroscopic sample has been derived by grouping the sources in bins of width @xmath33 , according to the expression @xmath34 where @xmath35 is the number of objects with luminosities between @xmath36 and @xmath37 and @xmath38 is the maximum comoving volume within which an object could have been detected above the radio - flux and magnitude limit of the survey .
we set @xmath39 mjy for the radio - flux limit , while @xmath1 is the chosen magnitude limit and the results for the local lf have been corrected for incompleteness effects coming from both the radio and the spectroscopic surveys .
+ this was done for each population taken individually .
fig.2 shows the results for the early - type ( lower panel ) , early plus e+agn ( middle panel ) and late - type+starburst ( top panel ) galaxies .
the dashed line in the bottom and middle panels of the figure is the predicted lf for steep spectrum fr i+fr ii sources as given by dunlop & peacock ( 1990 ) under the assumption of pure luminosity evolution . in this case
the agreement with the data is very good down to powers @xmath40 , especially if one includes all the objects which show spectra typical of early - type galaxies , regardless the presence of emission lines of agn origin .
the lf for late - type galaxies and starbursts is shown in the top panel of fig.2 , and features a break at about @xmath41 , beyond which the contribution of this class of objects becomes rapidly negligible .
when it comes to a comparison between observed and predicted lf for this latter case , one has that a good description of the data is provided by the rowan - robinson et al .
( 1993 ) model ( dashed line in the top panel of fig.2 ) , which can correctly reproduce both the broken power - law behaviour and the break luminosity , therefore supporting the assumption of these authors for late - type radio galaxies to be identified with the population of dusty spirals and starbursts observed at 60 @xmath42 m . however , the rowan - robinson et al .
( 1993 ) predictions result in an over - estimate of the number of faint ( i.e. @xmath43 ) radio sources , while the best fit is provided by assuming a shallower faint - end slope for the local lf , as illustrated by the dotted line in fig.2 ( see magliocchetti et al .
2001 for more details ) .
becker r.h . ,
white r.l . , helfand d.j . , 1995 , apj , 450 , 559 dunlop j.s . , peacock j.a . , 1990 , mnras , 247 , 19 maddox s.j . , 1998 , in `
large - scale structure : tracks and traces ' , proc .
12th postdam cosmogony workshop , world scientific , 91 ( astro - ph/9711015 ) magliocchetti m. , maddox s.j . , 2001 , to appear in mnras , astro - ph/0106429 magliocchetti m. , maddox s.j .
, + 2df team , 2001 , submitted to mnras , astro - ph/0106430 rowan - robinson m. , benn c. r. , lawrence a. , mcmahon r.g . , broadhurst t.j . , 1993 , mnras , 23 , 123 | we use redshift determinations and spectral analysis of galaxies in the 2df galaxy redshift survey to study the properties of local radio sources with @xmath0 mjy .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
an understanding of the nature and distribution of matter on small ( @xmath4 1 mpc ) scales is essential to modern astrophysics .
measuring the shapes of galaxy mass profiles reveals the presence of dark matter ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and its interactions with baryonic matter ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
separating total mass into luminous and dark components can constrain cosmological parameters ( e.g. the baryon fraction @xmath5 ) and place estimates on the efficiency of star - formation in galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
observing changes in any of these quantities over cosmological time is a key component in studying galaxy evolution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
while there are many techniques capable of measuring mass on small scales , gravitational lensing stands out as an especially powerful choice . unlike other methods , lensing directly measures
a total ( baryonic + dark matter ) mass without requiring this mass to be luminous or in any specific dynamical state .
furthermore , a lensing analysis is not limited to the local universe , but rather can be applied to systems located over a wide range of cosmological distances .
strong gravitational lensing , in particular , can provide a wealth of information about the nature of galaxies ( e.g. , @xcite and references therein ) . with typical image separations of @xmath6 1 arcsec
, galaxy - scale strong lenses produce mass estimates close to the centre of galaxies , giving information about both the baryon - dominated luminous core and the inner regions of the dark matter halo .
this relatively small angular size ( corresponding to physical scales of 5 - 10 kpc at typical lensing - galaxy redshifts ) decreases the probability of foreground interlopers contaminating the line of sight a problem that can bias mass estimates obtained from group and cluster - scale strong lenses , and can strongly dilute the signal measured from weak lensing suggesting that the observed lensing signal will be dominated by the mass of the lensing galaxy . in this paper
, we investigate the galaxy - scale gravitational lens b1938 + 666 .
first discovered as part of the jodrell bank very large array astrometric survey ( jvas ; @xcite ) , initial radio observations of b1938 + 666 showed a quadruply - imaged background source , configured into a partial einstein ring , along with a second , doubly - imaged component @xcite .
follow - up imaging in the near - infrared ( nir ) and optical uncovered a bright red object that was thought to be the galaxy lensing the radio emission @xcite .
this emission was later shown by nir _ hubble space telescope _
( _ hst _ ) imaging to be composed of the light from both the lensing galaxy and a complete einstein ring of the background source galaxy @xcite . @xcite measured a redshift of @xmath7 0.881 for the lensing galaxy from optical spectroscopy , while @xcite determined a redshift of @xmath8 2.059 for the source from co observations . as
b1938 + 666 has an einstein ring , its lensing mass model can be determined to high precision .
@xcite showed that , when compared with two - image quasar or arc lens systems , constraints from an einstein ring can be used to break degeneracies between the monopole moment of the gravitational potential and higher order terms .
removing these degeneracies allows for a more robust measurement of the slope of the lensing galaxy s mass profile and , when combined with time delay information , can provide an unambiguous measurement of the hubble constant ( e.g. @xcite ) . at the same time , with so many constraints placed on the model , einstein ring lenses are sensitive to perturbations from the smooth gravitational potential of the lens galaxy @xcite .
satellite galaxies and dark matter sub - haloes that orbit the main lensing galaxy can give rise to these perturbations , and while these objects are often too faint to be directly observed , gravitational lensing can be used to detect them indirectly through their mass signatures . in this way ,
careful scrutiny of einstein ring lenses ( or those with extended gravitational arcs ) provides one of the best opportunities to reveal the presence of , and constrain the properties of , extragalactic substructure @xcite .
this would serve as a direct test of the @xmath9cdm numerical simulations that predict this substructure ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and would thus advance our understanding into the nature of galaxy formation and evolution . for the reasons described above , we choose to study b1938 + 666 using high - resolution , nir adaptive optics ( ao ) as part of the strong lensing at high angular resolution program ( sharp ) .
this new project aims to obtain high resolution images of lens systems to study the mass distributions of lensing galaxies between 0.3 @xmath10 1 to an unprecedented detail with state - of - the - art imaging and lens modelling techniques .
in this first paper , we focus on a comparison between imaging data taken with space- and ground - based observatories , in particular , we determine the relative benefits for lens modelling
. we also present a new smooth mass model for our test system b1938 + 666 that is used to determine the properties of the lens galaxy and give a robust estimate of the magnification of the nir component of the background galaxy .
this paper is organized as follows . in section
[ sharp ] , we describe the wider goals of the sharp survey since this is the first paper in the series . in section [ obs ] ,
we briefly describe the new and archival multi - wavelength imaging data for b1938 + 666 , new ir spectroscopy of the lensed source and the techniques used to reduce them . in section
[ results ] , we present the results of lens modelling .
we discuss our results in section [ discussion ] . finally , we summarize and conclude in section [ conclusions ] .
measurements describing the relative fraction of substructure within the b1938 + 666 lensing galaxy are presented in a companion paper @xcite . throughout this paper ,
we assume a cosmological model of @xmath11 kms@xmath12 mpc@xmath12 , @xmath13 , and @xmath14 .
all magnitudes presented in this work are ab magnitudes .
the image separations seen in galaxy - scale lenses are typically on the order of an arcsecond .
thus , observing these systems with high angular resolution instruments can provide unambiguous estimates of individual image positions and magnitudes , information that is critical when measuring mass distributions with gravitational lensing .
furthermore , any improvements in angular resolution compared to traditional ground - based imaging will provide enhanced detectability of faint objects with small angular extent , and improved sensitivity to small - scale perturbations of the surface brightness of lensed extended emission . at optical and nir wavelengths , there are currently two techniques for obtaining the necessary data : space - based imaging with _
hst _ and ground - based ao imaging .
the sharp survey utilizes both approaches in order to build up a statistically significant sample of lens systems for which deep high - resolution imaging has been obtained .
a particular focus of the sharp survey is to detect and measure the mass of substructures associated with the lensing galaxies , without regard to whether the substructures are luminous or dark .
numerical simulations of galaxy formation predict a large amount of substructure for a galaxy - mass halo , with a mass fraction of @xmath15 510 per cent of the total halo mass contained in substructures with masses between 4 @xmath16 10@xmath17 and 4 @xmath18 10@xmath19 @xcite .
furthermore , the simulations converge on a substructure mass function of @xmath20 , where @xmath211.9 @xmath22 0.1 ( e.g. , * ? ? ? * ) .
@xcite have shown that a bayesian analysis of a sample of lenses that have been surveyed for substructure down to some mass threshold can provide meaningful constraints on @xmath23 and the slope of the mass function , @xmath24 .
given a sample of @xmath630 lens systems , a mass detection threshold of 3 @xmath18 10@xmath25 or better , and a reasonable prior on @xmath24 , good constraints on @xmath23 are obtained .
note that even the non - detection of substructure down to the mass limit provides useful information for constraining @xmath26 and @xmath24 , especially as the mass detection limit becomes smaller .
we are therefore pursuing two complementary methods for detecting substructure in lens systems , both of which utilize deep , high - resolution imaging .
the first is to detect luminous satellites directly , thus fixing the location of the substructure .
the substructure can therefore be included in the mass model of the lens and its mass can be determined ( e.g. , * ? ? ?
this method is most effective for lens systems with either four lensed images or extended lensed emission .
the other approach is to detect substructures through their gravitational effects on extended lensed emission ( e.g. , * ? ? ?
thus , the sharp sample consists of lens systems with quadruply imaged quasar lenses and lenses with partial or complete einstein rings .
results on two four - image lenses have been reported by @xcite and @xcite . here
we report the results for b1938 + 666 , which has an almost complete einstein ring .
a detailed description of the sharp sample will be presented by fassnacht et al .
( in preparation ) .
we observed the b1938 + 666 system on ut 2010 june 29 and 30 with the nirc2 camera on the keck - ii telescope , using the laser guide star ( lgs ) ao system .
we used the narrow camera , with a field - of - view of 10 arcsec @xmath18 10 arcsec , and a pixel scale of 0.01 arcsec .
the tip - tilt correction was obtained through simultaneous observation of a magnitude r = 15 star at a distance of 18 arcsec from the lens .
we observed the system using both the h and k@xmath27 filters .
details of the observations are given in table [ tbl : obs ] .
.a summary of the b1938 + 666 optical and infrared imaging observations .
[ cols="^,^,^,^,>",options="header " , ] [ tbl : modeling_results ] +
we now discuss our results .
first , we give a comparison of modelling the gravitational lens system using high angular resolution data from space- and ground - based telescopes .
we then discuss the properties of the lensing galaxy and the magnification of the background source that we infer from our model .
finally , we investigate the possibility of there being a second gravitationally lensed source for this system .
the precision and accuracy of mass model constraints from extended sources on strong gravitational lens galaxies is set by three conditions : ( i ) the number of independent resolution elements across the lensed images ( and their multiplicity ) , ( ii ) the average snr of the lensed - image surface brightness distribution and ( iii ) the level of surface brightness structure in the lensed source .
while the first two conditions set the level of information contained in the lensed images and how well , in principle , the mass model of the lens galaxy can be constrained , the third condition determines the level of covariance ( i.e. , degeneracy ) in the mass model , where more structured sources in general lead to a lower level of covariance between model parameters .
one of the goals of this paper has been to illustrate points ( i ) and ( ii ) by comparing the results of high snr _ hst _ f160w - band data with both lower resolution and pixel - sampling , and lower snr keck ao imaging data with higher resolution and sampling .
we implicitly assume that the intrinsic source structure in the f160w- , h- , and k@xmath27-bands is very much correlated . despite the fact that the results of the modelling of the three very different data sets are promisingly similar and the errors are small , we do note that there are differences .
in particular , there are discrepancies between the axis ratio and position angle ( pa ) of the lens mass distribution , and the external shear strength and pa ( see fig .
[ fig : posterior ] ) .
some tension between these quantities can be expected , as they are strongly covariant ( i.e. , the external shear mimics the flattening of the lens potential / mass - distribution ) and lack of information combined with errors on the data can lead to biases in the maximum - likelihood solutions and the posterior probability distributions of individual parameters .
this is especially true for b1938 + 666 , where a nearly circular mass distribution and largely featureless einstein ring conspire to add uncertainty to the lens model .
specifically , the circularly symmetric mass distribution ( coupled with nearly coaxial foreground and background galaxies ) makes it difficult to constrain the lensing galaxy s mass slope increasing the covariances between other parameters while the smooth light distribution lacks the contrast needed to differentiate between small - scale variations on the model .
thus , even moderate variations in parameter space can leave the lensed light distribution relatively unchanged , allowing significantly different models to fit the data equally well .
looking at the marginalized probability distributions in fig .
[ fig : posterior ] it is obvious that there is a strong covariance between the lens strength ( @xmath28 ) and mass slope ( @xmath29 ) , especially in the h - band and nicmos models .
this is why model discrepancies are the strongest between these data sets .
however , we note that these degeneracies , and the parameter uncertainties in general , are much smaller from the higher - resolution ( but lower snr ) keck ao data .
this implies that at least for some systems , ground - based ao data can `` out - perform '' space - based data in precision and most likely also in accuracy due to the better sampled lens image structure .
of course , this does not suggest that higher snr data is unnecessary : of the two ao - based models , the k@xmath27-band data ( with its relatively higher snr ) provides much tighter constraints on individual parameters than the h - band model , and indeed , even breaks the mass - slope degeneracy found in the h - band and nicmos models .
rather , the comparison simply shows that , between snr and image resolution , resolution seems to dominate any limitations on model precision .
one has to be careful , though , when attempting to generalize this conclusion to other lens systems , because the result depends critically on the complex interplay between snr , spatial resolution , and source structure .
it could very well be that lens systems with highly - structured sources ( e.g. , space - based b- or u - band data , where star formation could be strong and the sources could be more structured ) could out - perform the higher resolution k@xmath27-band ao data .
this interplay is currently under study . in cases where there is similar data , and
a bright star is available , however , our results show that ground - based ao data can perform significantly better than their space - based _ hst _ data sets in constraining lens models .
as we show in the companion paper , the higher resolution provided by ao imaging was also very powerful in the discovery of a low - mass substructure in this lens system @xcite .
the b1938 + 666 lensing galaxy is clearly seen in the optical and nir data sets ( fig .
[ fig : imgs ] ) , and its surface brightness distribution is well - modelled by a single elliptical srsic profile ( see section [ sb - model ] ) . excluding the hst v - band photometry
, we find the galaxy to have i@xmath30h and h@xmath30k@xmath27 colours of 1.9 and 0.4 magnitudes , respectively , after galactic reddening corrections .
we model these colours with the @xcite stellar population code and , assuming a solar metallicity and no dust , find consistency with an old ( @xmath31 4 gyr ) stellar population .
the foreground galaxy s isophotal regularity , colours , and absorption line spectrum measured by @xcite suggest that the lens has an early - type morphology ; this is consistent with the previous analysis of @xcite based only on the nicmos f160w - band imaging .
we use the @xcite 4 gyr stellar population model to compute the v - band absolute magnitude and to estimate the stellar mass within the einstein radius .
the luminosity is found to be @xmath32 3 @xmath18 10@xmath33 and the stellar template has a v - band stellar mass - to - light ratio of 1.54 , implying a stellar mass within the einstein radius of @xmath34 4.7 @xmath18 10@xmath35 if a chabrier initial mass function ( imf ) is assumed .
we can compare this to the _ total _ mass within the einstein radius from lensing , and we find a dark - matter mass fraction of @xmath36 0.55 within the einstein radius .
if we instead use a salpeter imf to describe the stellar population , the stellar mass within the einstein radius becomes @xmath34 8.2 @xmath18 10@xmath35 , leading to a dark - matter mass fraction of @xmath36 0.2 .
ideally , we would like to compute the stellar - to - total mass fraction within an aperture physically associated with the lensing galaxy .
therefore , we use two methods to estimate the effective radius from the more robustly measured lensing data .
first , we employ the relationship between the power - law density slope and effective radius found by @xcite and correct for early - type galaxy growth rates ( e.g. , * ? ? ?
there are several significant caveats , including evolution of the power - law slope ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and large uncertainties on these relationships ; nevertheless , we find that the effective radius is consistent with being the same size as the einstein radius .
motivated by this , we check if the assumption that the effective radius of 0.45 arcsec ( i.e. , equivalent to the einstein radius ) is consistent with measurements of the fundamental plane .
we use the velocity dispersion inferred from the lensing model as a proxy for the stellar velocity dispersion and apply a passive evolution correction to the luminosity as determined by the @xcite 4 gyr template .
we find that , again correcting for size evolution , our assumption that the effective radius and einstein radius are co - incident yields consistency with the fundamental plane presented by @xcite .
the gravitational lens models also provide an estimate of the source surface brightness distribution , after correcting for the lensing effect . from the pixelated reconstruction , we find that the source is composed of two components at nir wavelengths ( in agreement with the parametrized source distribution discussed in section [ sb - model ] ) .
one component is a high surface brightness region with a projected size of @xmath6 0.8 kpc , the second component is more extended , up to @xmath61.6 kpc in projected size , and has a lower surface brightness .
we defer any detailed discussion about the nature of the nir source reconstruction especially with respect to the agn and molecular gas components traced by the emission at radio wavelengths to a follow - up paper . to estimate the total magnification of the nir source from the gravitational lens models , we compare the model emission found in the lensing plane to that found in the source plane .
we use the _ hst _ nicmos data set for this calculation because these data have the most robust measurement of the extended light distribution of the source galaxy ; the ao data sets will have some light artificially pushed into an extended envelope ( thus leading to a greater chance of confusion between source and lens galaxy light ) due to the fact that ao strehl ratios are less than unity .
additionally , the brightness of the sky background at nir wavelengths can hide faint , extended emission at the outer edge of the source galaxy , leading to a biased magnification estimate . overall
, we find that the total magnification factor of the nir emission is @xmath6 13 .
we note that a magnification of 176 has been previously reported for b1938 + 666 from a simple point - source model ( see @xcite for some details ) .
this larger magnification was used by @xcite for their analysis of the molecular gas properties of b1938 + 666 , and they found that the co ( 32 ) line intensity for b1938 + 666 , and hence the molecular gas mass , was about an order of magnitude lower than for other quasars at a similar epoch .
this large discrepancy is almost certainly due to the magnification of the gas emission being over - estimated . however , although we find a smaller magnification for the nir emission region for this source , some caution should be taken when applying this value to other wavebands , particularly if the emission has a different size , or position with respect to the lensing caustics produced by the lens ; only spatially resolved imaging and a good lens model can give a robust estimate of the magnification for any particular emission region of a gravitationally lensed source . in @xcite
we presented the detection of a low - mass substructure in the b1938 + 666 system from the distorting effect the substructure has on the lensed arc .
we do not find any evidence of luminous , and probably more massive , substructure in the residual images for the system , which is consistent with the @xcite analysis . to determine an upper limit on the brightness of any luminous substructure
, we add a series of simulated point sources between 24.5 @xmath37 28 to an image of the lens system from which both the lens galaxy and the einstein ring emission have been subtracted . for our point source model
, we choose the empirical psf star used in the lens modelling , as this would accurately represent how a point source object would appear in the image .
one example simulation is presented in fig .
[ fig : psim ] .
we choose a point source model over other models such as gaussian or srsic profiles because we have no inherent knowledge about the morphologies of these objects , and because the typical small size of satellites relative to a parent galaxy , coupled with the high redshift of the b1938 + 666 lensing galaxy , suggests that these objects would be likely unresolved in our data .
additionally , since a point source object has a higher surface brightness than an extended object of the same magnitude , a point - source limiting magnitude represents a robust limit to the lowest luminosity that can possibly be observed , placing a hard limit on the luminous properties of any substructure that is independent of morphology .
we generate a total of 1000 simulated images , randomizing both the magnitudes and positions of the point sources , and then use sextractor @xcite to detect the added satellites .
we consider a point source to be detected if sextractor locates an object within 5 pixels and 1.5 magnitudes of the actual position and magnitude of the simulated object .
compiling the results into magnitude bins , we find that we are able to recover the majority ( i.e. @xmath38 50 per cent ) of the simulated objects down to an apparent magnitude of k@xmath39 26.1 , which we take to be our detection limit ( see fig . [
fig : plimit ] ) . converting this value to the rest - frame v - band corresponds to an object that has an absolute magnitude of @xmath4016.2 , which is fainter than the magellanic clouds ( @xmath4118.5 , @xmath4217.1 ) , but approximately three times brighter than the sagittarius dwarf satellite ( @xmath4315 ; @xcite ) .
26.1 bin ( represented by the dashed line ) .
we therefore treat this as our limiting magnitude.,width=302 ] there are two faint arclet structures that can be seen just outside the einstein ring on the eastern side of the k@xmath27 image ( fig .
[ fig : imgs ] ; bottom middle panel ) . to enhance the signal of these arclets
, we apply a gaussian filter ( @xmath44 6.5 pixels ) to smooth the data , followed by a laplacian filter to increase the contrast around bright flux peaks .
[ fig : arcs ] shows a high - contrast image of the original data and an image showing the results of applying gaussian and laplacian filters . while the two arcs are barely detectable above the noise in the unprocessed data , the arcs are clearly visible in the filtered image .
as the arclets are so faint in the original k@xmath27-band data , we do not formally consider them in our lens model .
however , taking into account the enhanced image , it is possible given their positions and orientations relative to the lensing galaxy that these arclets are actually multiple images of a second background source .
there are three possible scenarios : the arclets are not being lensed ; the arclets are images of a strongly - lensed source that is associated with the primary ring ( e.g. , an additional component of the galaxy being lensed into the ring , a nearby galaxy , or more images of the ring
although this would lead to an extremely peculiar image configuration that we do not expect to arise from well - behaved lensing mass distributions ) ; or the arclets are images of a second source being strongly lensed but at a completely different redshift than the einstein ring galaxy .
we can investigate these scenarios by using the lens models that have been fitted to the einstein ring .
we first use our best lens model to cast the emission from the einstein ring and the two arclets back to the source plane .
as expected , the arclets do not fall on the einstein ring source .
however , the two arclets also are not mapped to a coincident location on the source plane , indicating that if they are at the same redshift as the einstein ring galaxy they are not multiple images of the same source .
the possibility that the arclets correspond to images of a strongly - lensed source at a different redshift is intriguing ; such double - plane lens systems provide very tight constraints to the mass - density slope of the lens and can give useful constraints on cosmological parameters ( see @xcite for details ) . to investigate this possibility
we must rescale the lens strength of our mass model , where the scaling is directly proportional to the ratio of angular diameter distances between the lens and the two sources .
we find that the two arclets can be mapped to approximately the same location in the source plane if the lens strength is increased by a factor of @xmath6 1.85 .
such a large scaling factor requires an unrealistically high redshift for the source , although two caveats exist that could mitigate this concern : the galaxy being lensed into the einstein ring can contribute to the lensing , and the mass profile could deviate from the central power law that was found from the einstein ring fit . in any case , some estimate of the redshifts of the two potential gravitational arcs will need to be obtained to determine their nature .
this will likely require extensive followup with deep , high - resolution spectroscopy . while this is difficult with current telescopes ( especially because arclet b is so faint ) , future instruments such as the _ james webb space telescope _
( _ jwst _ ) , the thirty meter telescope ( tmt ) , or the european - extremely large telescope ( e - elt ) should provide the resolution and light - gathering power necessary to achieve this goal .
we have presented a new mass model for the gravitational lens system b1938 + 666 , using a grid - based bayesian reconstruction technique on high resolution ground - based ao and space - based _ hst _ data , as part of the strong - lensing at high angular resolution program ( sharp ) .
we find that the smooth component of the lensing galaxy s mass profile is well - fitted by a ( nearly isothermal ) power - law distribution , while its light profile ( and that of the source galaxy ) can be described by srsic components .
a more in - depth analysis of the mass model , characterizing the amount of substructure present in the system , is presented in a companion paper to this work @xcite .
the model is consistent over three independent nir data sets due largely to the constraints provided by a bright einstein ring and agrees well with previously reported results . when compared to models derived from traditional ground - based imaging , though , the high - resolution models are significantly more precise .
the relative improvement in precision varies from data set to data set .
thus , by generating both ao- and _ hst_-based models , we have provided a quantitative comparison between instruments .
overall , we find that the uncertainties on the model parameters derived from the ao data sets are smaller than those measured from the _ hst _ , by as much as an order of magnitude .
this suggests that ao data are better at constraining lens models than equivalent _
hst _ data ( at least in cases where the lens and source galaxies are red and have reasonably smooth light profiles ) , and therefore , that ao observations of lens systems can lead to a better description of those systems mass distributions .
however , confirmation of this result will require a much larger data set and will be explored in a future paper ( sharp ii ; fassnacht et al .
, in prep ) .
djl and cdf acknowledge support from nsf - ast-0909119 .
the centre for all - sky astrophysics is an australian research council centre of excellence , funded by grant ce110001020 .
sv is supported by a pappalardo fellowship at the massachusetts institute of technology and is grateful to the helena klyuver female visitor program for funding her stay at astron , during which part of this work was carried out .
lvek is supported ( in part ) through an nwo - vidi program subsidy ( project number 639.042.505 ) .
all of the authors would like to extend thanks to astron for the warm hospitality ( and financial support ) during the sharp workshop , held in dwingeloo , the netherlands .
this research was supported in part by the national science foundation under grant no .
nsf - phy11 - 25915 .
the data presented herein were obtained at the w. m. keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w. m. keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
the results present herein were also based on observations collected with the nasa / esa _ hst _ , obtained at stsci , which is operated by aura , under nasa contract nas5 - 26555 .
auger m. w. , treu t. , bolton a. s. , gavazzi r. , koopmans l. v. e. , marshall p. j. , bundy k. , moustakas l. a. , 2009 , apj , 705 , 1099 auger m. w. , treu t. , gavazzi r. , bolton a. s. , koopmans l. v. e. , marshall p. j. , 2010 , apj , 721 , l163 auger m. w. , treu t. , brewer b. j. , marshall p. j. , 2011 , mnras , 411 , l6 barkana r. , 1998 , apj , 502 , 531 barvainis r. , ivison r. , 2002 , apj , 571 , 712 behroozi p. s. , conroy c. , wechsler r. h. , 2010 , apj , 717 , 379 bertin e. , arnouts s. , 1996 , a&as , 117 , 393 blumenthal g. r. , faber s. m. , flores r. , primack j. r. , 1986 , apj , 301 , 27 bolton a. s. , burles s. , koopmans l. v. e. , treu t. , moustakas l. a. , 2006 , apj , 638 , 703 bolton a. s. et al . , 2012 ,
apjl , submitted ( arxiv:1201.2988 ) bosma a. , 1999 , aspc , 182 , 339 browne i. w. a. , wilkinson p. n. , patnaik a. r. , wrobel j. m. , 1998 , mnras , 293 , 257 bruzual g. , charlot s. , 2003 , mnras , 344 , 1000 conroy c. et al . , 2007 , apj , 654 , 153 de blok w. j. g. , mcgaugh s. s. , 1997 , mnras , 290 , 533 diemand j. , kuhlen m. , madau p. , 2007 ,
apj , 667 , 859 diemand j. , kuhlen m. , madau p. , zemp m. , moore b. , potter d. , stadel j. , 2008 , nat , 454 , 735 dutton a. a. , et al . , 2011 , mnras , 417 , 1621 feroz f. , hobson m. p. , 2008 ,
mnras , 384 , 449 fukugita m. , hogan c. j. , peebles p. j. e. , 1998 , apj , 503 , 518 gavazzi r. , treu t. , rhodes j. d. , koopmans l. v. e. , bolton a. s. , burles s. , massey r. j. , moustakas l. a. , 2007 , apj , 667 , 176 gavazzi r. , treu t. , koopmans l. v. e. , bolton a. s. , moustakas l. a. , burles s. , marshall p. j. , 2008 , apj , 677 , 1046 gnedin o. y. , kravtsov a. v. , klypin a. a. , nagai d. , 2004 , apj , 616 , 16 heymans c. et al . , 2006 , mnras , 371 , l60 hoekstra h. , yee h. k. c. , gladders m. d. , 2004 , apj , 606 , 67 king l. j. , browne i. w. a. , muxlow t. w. b. , narasimha d. , patnaik a. r. , porcas r. w. , wilkinson p. n. , 1997 , mnras , 289 , 450 king l. j. et al . , 1998 , mnras , 295 , l41 kochanek c. s. , 2006 , glsw.conf , 91 kochanek c. s. , keeton c. r. , mcleod b. a. , 2001 , apj , 547 , 50 koopmans l. v. e. , 2005 , mnras , 363 , 1136 lagattuta d. j. et al . , 2010 , apj , 716 , 1579 lagattuta d. j. , auger m. w. , fassnacht c. d. , 2010 , apj , 716 , l185 mandelbaum r. , seljak u. , kauffmann g. , hirata c. m. , brinkmann j. , 2006 , mnras , 368 , 715 mckean j. p. et al . , 2007 , mnras , 378 , 109 more a. , mckean j. p. , more s. , porcas r. w. , koopmans l. v. e. , garrett m. a. , 2009 , mnras , 394 , 174 mclean i. s. , et al . , 1998 , spie , 3354 , 566 napolitano n. r. , romanowsky a. j. , tortora c. , 2010 , mnras , 405 , 2351 newman a. b. , ellis r. s. , treu t. , bundy k. , 2010 , apj , 717 , l103 patnaik a. r. , browne i. w. a. , wilkinson p. n. , wrobel j. m. , 1992 , mnras , 254 , 655 riechers d. a. , 2011 , apj , 730 , 108 rhoads j. e. , malhotra s. , kundic t. , 1996 , aj , 111 , 642 rubin v. c. , peterson c. j. , ford w. k. jr . , 1980 , apj
, 239 , 50 ruff a. j. , gavazzi r. , marshall p. j. , treu t. , auger m. w. , brault f. , 2011 , apj , 727 , 96 schulz a. e. , mandelbaum r. , padmanabhan n. , 2010 , mnras , 408 , 1463 srsic j. l. , 1963 , baaa , 6 , 41 springel v. et al . , 2008 , mnras , 391 , 1685 suyu s. h. , halkola a. , 2010 , a&a , 524 , a94 suyu s. h. , marshall p. j. , auger m. w. , hilbert s. , blandford r. d. , koopmans l. v. e. , fassnacht c. d. , treu t. , 2010 , mnras , 711 , 201 suyu s. h. , et al . , 2012 ,
apj , 750 , 10 tollerud e. j. , bullock j. s. , strigari l. e. , willman b. , 2008 , apj , 688 , 277 tonry j. l. , kochanek c. s. , 2000 , aj , 119 , 1078 van albada t. s. , sancisi r. , 1986 , rspta , 320 , 447 vegetti s. , koopmans l. v. e. , 2009a , mnras , 392 , 945 vegetti s. , koopmans l. v. e. , 2009b , mnras , 400 , 1583 vegetti s. , czoske o. , koopmans l. v. e. , 2010 , mnras , 407 , 225 vegetti s. , koopmans l. v. e. , bolton a. , treu t. , gavazzi r. , 2010 , mnras , 408 , 1969 vegetti s. , lagattuta d. j. , mckean j. p. , auger m. w. , fassnacht c. d. ,
koopmans l. v. e. , 2012 , nat , 481 , 341 wilkinson p. n. , browne i. w. a. , patnaik a. r. , wrobel j. m. , sorathia b. , 1998 , mnras , 300 , 790 | we present new mass models for the gravitational lens system b1938 + 666 , using multi - wavelength data acquired from keck adaptive optics ( ao ) and _ hubble space telescope _ ( _ hst _ ) observations .
these models are the first results from the strong - lensing at high angular resolution program ( sharp ) , a project designed to study known quadruple - image and einstein ring lenses using high - resolution imaging , in order to probe their mass distributions in unprecedented detail . here , we specifically highlight differences between ao- and _ hst_-derived lens models , finding that at least when the lens and source galaxies are both bright and red , and the system has a high degree of circular symmetry
ao - derived models place significantly tighter constraints on model parameters . using this improved precision ,
we infer important physical properties about the b1938 + 666 system , including the mass density slope of the lensing galaxy ( @xmath0 ) , the projected dark matter mass fraction within the einstein radius ( @xmath1 ) , and the total magnification factor of the source galaxy ( @xmath2 ) .
additionally , we measure an upper - limit constraint on luminous substructure ( @xmath3 ) , based on the non - detection of bright satellite galaxies in all data sets . finally , we utilize the improved image resolution of the ao data to reveal the presence of faint arcs outside of the primary einstein ring . the positions and orientations of these arcs raise the intriguing possibility that b1938 + 666 has a second source galaxy , located at a more distant redshift . however , future work is needed to verify this hypothesis .
galaxies : individual ( jvas b1938 + 666 ) gravitational lensing : strong |
a general predictions of any quantum theory of gravity is that there exists a minimal length below which no other length can be observed @xcite-@xcite . from perturbative string theory point of view @xcite
, such a minimal observable length is due to the fact that strings can not probe distances smaller than the sting size . in the scale of this minimal size ,
the quantum effects of gravitation become as important as the electroweak and strong interactions .
thus , in the study of high energy physics phenomena such as the very early universe or in the strong gravitational fields of a black hole , one can not neglect the effects of the existence of such a minimal length .
an important feature of the existence of a minimal length is the modification of the standard heisenberg commutation relation in the usual quantum mechanics @xcite .
such relations are known as the generalized uncertainty principle ( gup ) . in one dimension ,
the simplest form of such relations can be written as @xmath0 where @xmath1 and @xmath2 are positive and independent of @xmath3 and @xmath4 , but may in general depend on the expectation values @xmath5 and @xmath6 . the usual heisenberg commutation relation can be recovered in the limit @xmath7 . as is clear from equation ( [ a ] ) , this equation implies a minimum position uncertainty of @xmath8 , and hence @xmath1 must be related to the planck length . for a more general discussion on such deformed heisenberg algebras , especially in three dimensions ,
see @xcite .
now , it is possible to realize equation ( [ a ] ) from the following commutation relation between position and momentum operators @xmath9=i\hbar \left(1+\beta p^2\right),\ ] ] where we take @xmath10 .
more general cases of such commutation relations are studied in @xcite-@xcite . also various applications of the low energy effects of the modified heisenberg uncertainty relations
have been extensively studied , see for example @xcite-@xcite .
in this letter we consider a two dimensional minisuperspace of bianchi type i cosmology in the gup framework .
we shall see that the corresponding wheeler - dewitt ( wd ) equation is a fourth order differential equation .
although , in general we can not solve this equation exactly , we may obtain approximate analytical solutions in the limit of small scale factors _ i.e. _ , in the region of the validity of gup .
we then compare the resulting wave functions with ordinary quantum cosmology and with noncommutative quantum cosmology where the latter is discussed in @xcite .
let us consider a cosmological model in which the spacetime is assumed to be of bianchi type i whose metric can be written , working in units where @xmath11 , as @xmath12 where @xmath13 is the lapse function , @xmath14 is the scale factor of the universe and @xmath15 determine the anisotropic parameters @xmath16 and @xmath17 as follows @xmath18 to simplify the model we take @xmath19 , which is equivalent to a universe with two scale factors in the form @xmath20 the anisotropy in the above metric can achieved by introducing a large scale homogeneous magnetic field in a flat frw spacetime .
such a magnetic field results in a preferred direction in space along the direction of the field .
if we introduce a magnetic field which has only a @xmath21 component , the resulting metric can be written in the form ( [ e ] ) where there are equal scale factors in the transverse directions @xmath22 and @xmath23 and a different one , @xmath24 , in the longitudinal direction @xmath21 .
the dynamics of such a universe is considered in @xcite .
now , using the einstein - hilbert action @xmath25 where @xmath26 , @xmath27 and @xmath28 represent the determinant of the metric tensor , the scalar curvature and the cosmological constant respectively , we are led to the following lagrangian in the minisuperspace @xmath29 @xmath30 thus , with the choice of the harmonic time gauge @xmath31 @xcite , the hamiltonian can be written as @xmath32 let us now proceed to quantize the model . in ordinary canonical quantum cosmology , use of the usual commutation relations @xmath33=i\delta_{ij}$ ] , results in the well known representation @xmath34 , from which the wd equation can be constructed
. however , in the gup framework , as was mentioned in the introduction , the existence of a minimal observable length requires new commutation relations between position and momentum operators . in more than one dimension a natural generalization of equation ( [ b ] ) is defined by the following commutation relations @xcite @xmath35=i\left(\delta_{ij}+\beta \delta_{ij}p^2+\beta'p_ip_j\right),\ ] ] where @xmath36 and @xmath37 are considered as small quantities of first order . also , assuming that @xmath38=0,\ ] ] the commutation relations for the coordinates are obtained as @xmath39=i\frac{(2\beta-\beta')+(2\beta+\beta')\beta p^2}{1+\beta p^2}\left(p_ix_j - p_jx_i\right).\ ] ] as it is clear from the above expression , the coordinates do not commute .
this means that to construct the hilbert space representations , one can not work in position space .
it is therefore more convenient to work in momentum space , as is done in @xcite , @xcite and @xcite .
however , since in quantum cosmology the wave function of the universe in momentum space has no suitable interpretation , we restrict ourselves to the special case @xmath40 . as one
can see immediately from equation ( [ l ] ) , the coordinates commute to first order in @xmath1 and thus a coordinate representation can be defined .
now , it is easy to show that the following representation of the momentum operator in position space satisfies relations ( [ i ] ) and ( [ j ] ) ( with @xmath40 ) to first order in @xmath1 @xmath41 a comment on the above issue is that applying the gup to a curved background such as a cosmological model needs some modifications @xcite . here ,
since we apply the gup to the minisuperspace variables @xmath42 which correspond to a minkowskian metric , we can safely use the above expressions without any modifications .
let us focus attention on the study of the quantum cosmology of the model described by the hamiltonian ( [ h ] ) .
the corresponding commutation relations are as follows @xmath43=i\left(1+\beta
p^2 + 2\beta p_u^2\right),\hspace{.5cm}\left[v , p_v\right]=i\left(1+\beta p^2 + 2\beta p_v^2\right),\ ] ] @xmath44=\left[v , p_u\right]=2i\beta p_u p_v,\ ] ] @xmath45=\left[p_i , p_j\right]=0,\hspace{.5cm}x_i(i=1,2)=u , v , \hspace{.5cm}p_i(i=1,2)=p_u , p_v.\ ] ] as we have mentioned in the previous section , in the special case when @xmath40 , we have the following representations for @xmath46 and @xmath47 in the @xmath48 space which satisfy the commutation relations ( [ m])-([o ] ) @xmath49 now , using these representations for momenta in the hamiltonian constraint ( [ h ] ) , the wd equation can be written , up to first order in @xmath1 , as @xmath50 in the case of @xmath51 , the ordinary quantum cosmology of the model can be recovered and its eigenfunctions can be written in terms of bessel functions as follows @xcite @xmath52 @xmath53 the solutions of equation ( [ r ] ) are separable and may be written in the form @xmath54 , leading to @xmath55 @xmath56 where for having well - defined functions we use the positive separation constant @xmath57 in the case of a positive cosmological constant . equation ( [ u ] ) is a fourth order linear differential equation whose solutions can be written in the form of exponential functions @xmath58 where @xmath59 is the root of equation @xmath60 where @xmath61.\ ] ] to achieve the correct limit for @xmath62 , we take the negative sign in the above expression .
thus , up to first order in @xmath1 , the solution of equation ( [ u ] ) reads @xmath63 now , let us deal with equation ( [ v ] ) .
taking @xmath51 in this equation yields the ordinary wd equation where its solutions in the case of a positive @xmath28 are given by equation ( [ s ] ) . in the case
when @xmath64 , we can not solve equation ( [ v ] ) exactly , but we can provide an approximate method whose domain of validity is given by the solution of a second order differential equation . to this end , we note that the effects of @xmath1 become important at the planck scale , or in cosmology language in the very early universe , that is , when the scale factor is small , @xmath65 .
thus , if we substitute solution ( [ s ] ) in the @xmath1-term of equation ( [ v ] ) , we may obtain approximate analytical solutions in the region @xmath66 .
the limiting behavior of solution ( [ s ] ) in the region @xmath67 is @xcite @xmath68 and therefore its fourth derivative is @xmath69 substituting this result into equation ( [ v ] ) leads to the following equation @xmath70u=0,\ ] ] with solution , up to first order in @xmath1 , as @xmath71 for a negative cosmological constant , the above procedure leads to the solutions @xmath72 for @xmath73 and modified bessel function @xmath74 for @xmath75 .
thus , the eigenfunctions of the wd equation can be written as @xmath76 @xmath77we may now write the general solutions to the wd equation as a superposition of the eigenfunctions @xmath78where @xmath79 can be chosen as a shifted gaussian weight function @xmath80 .
ccc in a previous work @xcite , we obtained the solutions of the same problem in a noncommutative phase space with the following commutation relations @xmath81=i\theta,\hspace{.5cm}\left[u_{nc},p_u\right]= \left[v_{nc},p_v\right]=i,\hspace{.5cm}\left[p_u , p_v\right]=0,\ ] ] and showed that the eigenfunctions of the corresponding wd equation in such a space are given by @xmath82 @xmath83 as can be seen , the general solutions are again in the form of an expression like ( [ af ] ) .
although in the present study and with the choice @xmath40 , the phase space variables @xmath84 commute with each other , in general equation ( [ l ] ) shows that gup leads naturally to a noncommutative generalization of position space which may point to a close relationship between noncommutativity and gup .
the study of such phenomena as ir / uv mixing and non - locality , lorentz violation and new physics at very short distances in noncommutative @xcite-@xcite and gup @xcite frameworks , would pave the way for a more clear understanding of this relationship .
figures [ fig1 ] and [ fig2 ] show the square of wave functions in the context of an ordinary commutative phase space , noncommutative phase space and when the phase space variables obey the gup relations .
as is clear from these figures , in the ordinary commutative case ( @xmath85 ) , we have only one possible universe around a nonzero value of @xmath86 and @xmath87 , which means that the universe in this case approaches a flat frw one .
on the other hand we see that noncommutativity causes a shift in the minimum value of @xmath86 corresponding to the spatial volume .
the emergence of new peaks in the noncommutative wave packet may be interpreted as a representation of different quantum states that may communicate with each other through tunnelling .
this means that there are different possible universes ( states ) from which our present universe could have evolved and tunnelled , from one state to another ( see also @xcite ) .
such behavior also occurs in figure [ fig2 ] which shows the square of the gup wave function , showing that from the point of view adopted here , noncommutativity and gup may be considered as similar concepts .
however , there is an important difference , namely , that the noncommutative wave function not only peaks around @xmath87 , but appear symmetrically around a nonzero value of @xmath88 , which is the characteristic of an anisotropic universe . on the other hand , the gup wave function as is seen in the figure ,
has many peaks around the value @xmath87 and therefore from the point of predicting an isotropic universe the gup wave packet behaves like the ordinary commutative case .
in this paper we have studied the effects of the existence of a minimal length scale on the quantum states of a bianchi type i cosmology .
this phenomena yields a deformed heisenberg commutation relation between the position and momentum operators which are known as the gup .
although in more than one dimension , because of the existence of a noncommutative relation between the space operators it is not possible to represent the momentum operator in the position space , in the special case when the gup parameters obey the relation @xmath40 , the space operators commute and a coordinate representation of momenta can be defined . since a wave function describing the quantum state of the universe does not not have a suitable interpretation in momentum space , working in the above special case is crucial in our work . when the phase space variables obey the ordinary commutation relations we have seen that there is only one possible isotropic universe .
upon considering the gup and through finding the approximate analytical solutions of the wd equation in the limit of small scale factors , we have studied the corresponding quantum cosmology and seen that in the presence of the gup , the square of the wave function of the universe has several peaks .
this behavior , also occurring in the noncommutative quantum model studied in a previous work @xcite , may be related to different states in the early universe from which our present universe could have evolved .
although , both noncommutativity and gup predict many possible initial universes , this is not the case in predicting the isotropicity . in the case of a noncommutative cosmology
the universe behaves anisotropicaly while the gup predicts an isotropic universe .
99 d. j. gross and p. f. mende , _ nucl .
_ b * 303 * , ( 1988 ) 407 .
d. amati , m. ciafaloni and g. veneziano , _ phys .
_ b * 216 * ( 1989 ) 41 .
m. kato , _ phys .
_ b * 245 * ( 1990 ) 43 .
s. haro , _ j. high energy phys .
jhep * 10 * ( 1998 ) 023 . l. g. garay , _ int .
phys . _ a * 10 * ( 1995 ) 145 , ( gr - qc/9403008 ) .
k. konishi , g. paffuti and p. provero , _ phys .
_ b * 234 * ( 1990 ) 276 .
a. kempf , g. mangano and r. b. mann , _ phys .
_ d * 52 * ( 1995 ) 1108 , ( hep - th/9412167 ) .
a. kempf and g. mangano , _ phys .
d * 55 * ( 1997 ) 7909 , ( hep - th/9612084 ) . m. maggiore , _ phys .
_ b * 319 * ( 1993 ) 83 , + m. maggiore , _ phys .
_ d * 49 * ( 1994 ) 5182 , + a. kempf , _
* 35 * ( 1994 ) 4483 , + r. j. adler and d. i. santiago , _ mod .
lett . _ a * 14 * ( 1999 ) 1371 .
a. kempf , _ phys .
_ d * 63 * ( 2001 ) 083514 , + a. kempf and j. c. niemeyer , _ phys .
_ d * 64 * ( 2001 ) 103501 , + l. n. chang , d. minic , n. okamura and t. takeuchi , _ phys .
_ d * 65 * ( 2002 ) 125027 . h. hinvichsen and a. kempf , _ j. math
. phys . _ * 37 * ( 1996 ) 2121 .
f. brau , _ j. phys . _ a * 32 * ( 1999 ) 7691 , ( quant - ph/9905033 ) .
r. akhoury and y. p. yao , _ phys .
_ b * 572 * ( 2003 ) , ( hep - ph/0302108 ) .
s. hossenfelder , m. bleicher , s. hofmann , j. ruppert , s. scherer and h. stcker , _ phys .
_ b * 575 * ( 2003 ) 85 , ( hep - th/0305262 ) .
k. nozari and t. azizi , _ gen .
* 38 * ( 2006 ) 735 .
b. vakili , n. khosravi and h. r. sepangi , _ class .
quantum grav . _
* 24 * ( 2007 ) 931 , ( gr - qc/0701075 ) . e. j. king and p. coles , _ class .
quantum grav .
_ * 24 * ( 2007 ) 2061 .
u. bleyer , v. d. ivashchuk , v. n. melnikov and a. zhuk , _ nucl .
_ b * 429 * ( 1994 ) 177 , ( gr - qc/9405020 ) , + v. d. ivashchuk , v. n. melnikov and a. i. zhuk , _ nuov .
_ d * 104 * , ( 1989 ) 575 .
a. kempf , _
_ a * 30 * ( 1997 ) 2093 ( hep - th/9604045 ) , + i. dadic , l. jonke and s. meljanac , _ phys .
_ d * 67 * ( 2003 ) 087701 , ( hep - th/0210264 ) . s. f. hassan and m. s. sloth , _ nucl .
_ b * 674 * ( 2003 ) 434 , + m. v. battisti and g. montani , _ the big - bang singularity in the framework of a generalized uncertainty principle _ , ( gr - qc/0703025 ) , + f. scardigli , _ phys .
b * 452 * ( 1999 ) 39 , ( hep - th/9904025 ) , + f. scardigli and r. casadio , _ class .
quantum grav . _ * 20 * ( 2003 ) 3915 .
w. magnus , f. oberhettinger and r. p. soni , _ formulas and theorems for the special functions of mathematical physics _ , springer - verlag , berlin , 1966 .
s. minwalla , m. van raamsdonk and n. seiberg , _ j. high energy phys .
_ jhep * 02 * ( 2000 ) 020 , ( hep - th/9912072 ) . s. m. carroll , j. a. harvey , v. a. kostelesky , c. d. lane and t. okamoto , _ phys . rev
* 87 * ( 2001 ) 141601 , ( hep - th/0105082 ) , + c. e. carlson , c. d. carone and r. f. lebed , _ phys .
_ b * 518 * ( 2001 ) 201 , ( hep - ph/0107291 ) , + c. e. carlson , c. d. carone and r. f. lebed , _ phys .
_ b * 549 * ( 2002 ) 337 , ( hep - ph/0209077 ) .
r. j. szabo , _ phys .
* 387 * ( 2003 ) 207 , ( hep - th/0109162 ) .
m. r. douglas and n. a. nekrasov , _ rev .
* 73 * ( 2002 ) 977 , ( hep - th/0106048 ) .
i. bars , _ nonpertubative effects of extreme localization in noncommutative geometry _ , ( hep - th/0109132 ) .
a. w. peet and j. polchinski , _ phys .
_ d * 59 * ( 1999 ) 065011 . c. m. hull ,
_ j. high energy phys .
_ jhep * 07 * ( 1998 ) 021 , + v. balasubramanian , p. horava and d. minic , _ j. high energy phys .
_ jhep * 05 * ( 2001 ) 043 , + v. balasubramanian , j. de boer and d. minic , _ phys .
_ d * 65 * ( 2002 ) 123508 . h. garcia- compean , o. obregon and c. ramirez , _ phys .
_ * 88 * ( 2002 ) 161301 , ( hep - th/0107250 ) . | we study a quantum bianchi type i model in which the dynamical variables of the corresponding minisuperspace obey the generalized heisenberg algebra .
such a generalized uncertainty principle has its origin in the existence of a minimal length suggested by quantum gravity and sting theory .
we present approximate analytical solutions to the corresponding wheeler - dewitt equation in the limit where the scale factor of the universe is small and compare the results with the standard commutative and noncommutative quantum cosmology .
similarities and differences of these solutions are also discussed .
pacs numbers : 04.60.-m , 04.60.ds , 04.60.kz -2.5 cm |
the connections between cosmology and particle physics through dark datter ( dm ) have recently received special attention for defining the physics program at the tev frontier .
we foresee that the combination of data from satellites , direct dm searches , hadron and lepton colliders will provide a major breakthrough in our understanding of the nature of dark matter and its interactions in the early universe .
these expectations are supported by the fact that there are several extensions of the standard model ( sm ) , which include a new , stable , weakly - interacting massive particle , which may be responsible for the observed relic dm in the universe .
this particle should become accessible to particle colliders operating at the tev energy frontier , as well as to the next generations of direct dm search experiments .
the lhc collider will be first in providing data to address the question of whether one of these scenarios is indeed realised in nature .
if this is the case , it will also gather some quantitative information to be related to the relic dm density measured from the cosmic microwave background ( cmb ) spectra @xcite . however , it is understood that the lhc data will not be exhaustive in this respect .
first , it will not be possible to infer , in a model independent way , the relic density to an accuracy close to that already achieved by cmb observations .
furthermore , there exist classes of models of new physics which the lhc may not be able to disentangle and probe in sufficient details .
it is only with the measurements becoming available at an electron positron collider , operating at centre - of - mass energies of order of 1 tev , that we shall be able to determine the properties of the dm candidate particle and of the other particles participating in its interactions in the early universe , with sufficient accuracy to predict the dm relic density precisely . with these results in hand
, the comparison of the data from cmb experiments , direct dm searches and collider experiments would have striking consequences on our quantitative understanding of the nature and distribution of dark matter in the universe . in these years
preceding lhc operation , supersymmetry has emerged as the best motivated theory of new physics beyond the sm .
it solves a number of open problems intrinsic to the sm and , most important to our discussion , the conservation of r - parity introduces the lightest neutralino , @xmath2 , as a new stable , weakly interacting particle .
cmb data from the wmap satellite , and other astrophysical data , already set rather stringent bounds on the parameters of supersymmetry , if the lightest neutralino is responsible for saturating the amount of dm observed in the universe .
the recently released , five - year wmap data provide a determination of the dark matter density as @xmath3 = [email protected] @xcite . the potential of the lhc and of an @xmath5 linear collider operating at 0.5 tev and 1.0 tev , such as the international linear collider ( ilc ) , for determining the neutralino relic density , @xmath6 , in supersymmetry has been investigated in detail in @xcite .
that study selected a set of benchmark points , the so - called lcc points , representative of various supersymmetric scenarios and determined the @xmath6 probability density function by a scan of the full parameter space of the minimal supersymmetric extension of the sm ( mssm ) , by retaining those points compatible with the measurements available at the lhc and ilc , within their experimental accuracy . in this paper
we consider one of the supersymmetric scenarios defined in @xcite , for which the neutralino relic density is controlled by its annihilation rate through the cp - even heavy higgs pole @xmath7 , which in turn crucially depends on the value of the mass of the boson , @xmath8 .
we study the accuracy of the measurement of the relevant properties of the neutral heavy higgs boson @xmath9 : its mass , @xmath8 , width , @xmath10 and decay branching fractions as can be obtained from data collected in high luminosity @xmath5 collisions at centre - of - mass energy of 1 tev , using full simulation of the response of a realistic detector model and detailed event reconstruction .
we adopt the lcc-4 benchmark point of @xcite , which is defined in the reduced paramater space of the constrained mssm by @xmath11=380 gev , @xmath12=420 gev , @xmath13=53 , @xmath14=0 , @xmath15=+1 and @xmath16=178 gev .
we use isasugra 7.69 @xcite to compute the physical particle spectrum and we get @xmath8=419.4 gev , @xmath17=169.1 gev and @xmath18=195.5 gev .
these parameters correspond to a neutralino relic density of @xmath19 = 0.108 , as obtained by using the micromegas 2.0 program @xcite .
the @xmath20 process at @xmath1 = 1 tev has already been studied for lcc-4 using a parametric simulation @xcite .
we now perform a detailed study using geant-4-based full simulation @xcite of the detector response and reconstruct the physics objects using processors developed in the marlin framework @xcite and extend the analysis to both the @xmath21 and @xmath22 final states .
this study adopts the ldc detector concept , which is based on a large continuous gaseous tracker , a time projection chamber , surrounded by a highly granular siw calorimeter and complemented by a high resolution si vertex tracker .
the ldc detector concept is discussed in detail elsewhere@xcite , the design is optimised for achieving excellent parton energy measurements through the particle flow algorithm , and precise extrapolation of particle tracks to their production point .
both of these features are important to this analysis , which aims at suppressing backgrounds by exploiting the signature 4-@xmath23 and 2-@xmath23 + 2-@xmath24 final states of the signal , and requires good determinaton of energy and direction of hadronic jets to attain an optimal resolution on di - jet invariant mass .
signal events have been generated with pythia 6.205 @xcite + isasugra 7.69 , including beamstrahlung effects @xcite . at @xmath1
= 1 tev , the effective @xmath25 production cross section , accounting for beamstrahlung and initial state radiation , is 1.4 fb , br(@xmath26 ) = br(@xmath27 ) = 0.87 and br(@xmath28 ) = br(@xmath29 ) = 0.13 .
the main particle pair production backgrounds , @xmath30 , @xmath31 and @xmath32 , have been generated using pythia .
their cross sections , computed using comphep 4.4.0 @xcite , are 0.17 pb , 3.0 pb and 0.19 pb respectively .
the inclusive @xmath21 and @xmath22 production , after subtracting the contribution of the @xmath33 channel and requiring 200 gev @xmath34 600 gev , have cross sections of 0.63 fb and 0.28 fb respectively .
these processes have been generated at parton level using comphep and then hadronised with pythia .
we assume to operate the linear collider at @xmath1=1 tev for a total integrated luminosity of 2 ab@xmath35 , which corresponds to 5 years ( 1 yr = 10@xmath36 s ) of operation for a nominal luminosity of @xmath37 @xmath38 s@xmath35 .
a loose event preselection based on event variables has been applied after generation .
selected signal and background events have been passed through the full ldc simulation using the mokka 06 - 03 program @xcite , an ilc - specific implementation of geant-4 .
data are persisted using lcio @xcite collections and used as input for the subsequent reconstruction in marlin .
pattern recognition and track fit are performed first using monte carlo truth information ( `` mc truth patrec '' ) and , for signal events , also genuine full pattern recognition ( `` full patrec '' ) , using the fullldctracking package based on delphi experiment software @xcite .
the performances of these two approaches are compared .
the pandora v02 - 00 package is used for particle flow @xcite .
jet clustering is performed using the durham algorithm @xcite .
the jet energy resolution has been studied using a simulated sample of single @xmath23 jets in the energy range from 10 gev to 210 gev over a polar angle , @xmath39 .
we get @xmath40 , which is consistent with the ldc particle flow performance specifications .
jet flavour tagging is performed using the lcfivertex package , which developed the original zvtop tagger @xcite and feeds track and vertex topological information into a neural network to distinguish between @xmath23 , @xmath41 and light quark jets .
the di - jet mass resolution in the @xmath21 has been improved by performing a constrained kinematic fit .
we have ported the pufitc algorithm @xcite , developed for the delphi experiment at lep2 , into a dedicated marlin processor .
the algorithm adjusts the momenta of the jets given by @xmath42 where @xmath43 is the fitted momentum , @xmath44 is the measured momentum , @xmath45 and @xmath46 are unit vectors transverse to @xmath44 and to each other , and @xmath47 , @xmath23 and @xmath41 are free parameters in the fit .
the adjusted momenta satisfy a set of constraints while minimising the fit @xmath48 , given by @xmath49 @xmath50 , where @xmath51 is the expected energy loss parameter , @xmath52 is the energy spread parameter and @xmath53 , @xmath54 are the transverse momentum spread parameters . in this analysis , we impose the following constraints : @xmath55 and @xmath56 , where the last condition accounts for beamstrahlung along the beam axis , @xmath57 .
first we analyse the fully hadronic final state .
this provides with characteristics four @xmath23 jet , symmetric events .
the backgrounds can be significantly suppressed using @xmath23-tagging , event - shape and kinematic variables .
we require selected events to fulfill the following criteria : total recorded energy in the event @xmath58 700 gev , total transverse energy @xmath59350 gev , total number of reconstructed particles @xmath6080 , number of charged particles @xmath6130 , event thrust @xmath620.9 and @xmath630.002 , where @xmath64 is the 3 to 4 jet cross - over value of the jet clustering algorithm .
the distributions of some of these variables are shown in figure [ fig : cuts ] for backgrounds and signal , for which a comparison of the generator - level and reconstructed values is also given .
after event selection , particles are forced into four jets , which are arranged into two di - jet pairs , using the pairing which minimises the difference between the di - jet masses , @xmath65 .
the kinematic fit is performed and a cut applied on the resulting di - jet mass difference @xmath6650 gev to eliminate poorly reconstructed events .
both di - jet masses are required to satisfy @xmath67200 gev .
the event is required to have four @xmath23 jets , where a @xmath23 jet is determined by the following criteria : total jet multiplicity @xmath6010 , charged jet multiplicity @xmath685 , and b jet probability , @xmath69 , larger than 0.5 . at the chosen working point ,
an efficiency for @xmath23 jets of 0.79 is obtained , using `` mc truth patrec '' , with sufficient rejection of lighter quarks to effectively suppress the remaining non-@xmath23 backgrounds . by using `` full patrec '' without retraining the neural net ,
we measure a tagging efficiency of 0.72 per jet .
events , in which the incorrect jet pairing ( ijp ) is chosen , are considered as background . ]
the di - jet mass for signal @xmath70 events fulfilling the selection cuts has a gaussian resolution of 23 gev using tracks reconstructed with `` mc truth patrec '' and 27 gev using tracks from `` full patrec '' before the kinematic fit . after applying the kinematic fit
the di - jet mass resolutions become 13.7 gev and 13.8 gev , respectively after final selection , the sample of events with di - jet masses in the region 200 gev @xmath71 550 gev gives a selection efficiency for signal @xmath72 decays of [email protected] using tracks reconstructed with `` mc truth patrec '' and [email protected] using `` full patrec '' .
the difference is mostly caused by the observed drop in b - tagging efficiency .
the corresponding acceptance for @xmath30 , @xmath31 , @xmath32 and inclusive @xmath21 background events is @xmath73 , @xmath74 , @xmath75 and @xmath76 , respectively .
the resulting mass distribution is shown in figure [ fig : jjmass ] , which has two entries per event .
the signal is described by the convolution of two breit - wigner functions with a mass splitting of 1.4 gev , as predicted for the lcc-4 parameters , convoluted with a double gaussian resolution function .
the background is described by a third - order polynomial with coefficients determined on background only events .
the final fit function consists of a linear combination of the signal and background functions with four free parameters : @xmath77 , @xmath78 , and the weights of the signal and background functions .
we get @xmath77 = ( [email protected] ) gev and @xmath78 = ( [email protected] ) gev , where the quoted uncertainties are statistical only .
this result is remarkably close to that obtained in the earlier analysis , based on parametric detector simulation . using `` full patrec '' the uncertainties on the @xmath9 boson mass and width increase to 1.3 gev and 3.4 gev , respectively .
the mixed decay mode @xmath22 can be isolated by tagging a @xmath80 di - jet , consistent with originating from either a @xmath81 or a @xmath9 decay and analysing the remaining particles in the event .
we require the events to fullfill the following criteria : @xmath82400 gev , 200 gev@xmath83900 gev , 40@xmath84180 , 15@xmath85100 , event [email protected] , event sphericity @xmath860.1 and @xmath630.005 .
the event is forced to four jets of which two must be tagged as @xmath23 jets using the same criteria as above but the tighter requirement @xmath87 0.9 .
the invariant mass of the @xmath88 di - jet must satisfy 300 gev@xmath89600 gev , and that of the two remaining jets 250 gev@xmath90600 gev .
the angle between the two @xmath23 jets and the angle between the two un - tagged jets must satisfy -0.8@xmath910 .
the number of charged particles with energy greater than 5 gev which are not associated to either of the @xmath23 jets must not exceed six .
finally , @xmath24 tagging is performed .
we have developed an algorithm which outputs a linear discriminant variable @xmath92 based on the jet mass , the impact parameter of the leading track , and a variable , @xmath93 , which measures the jet energy deposited in an annulus around the jet direction .
at least one of the two non-@xmath23 jets must be tagged as a @xmath24 jet , where a @xmath24 jet must have less than four energetic charged particles and must satisfy @xmath940.8 . to distinguish between signal @xmath95 and @xmath22 decays , a discriminant variable @xmath96 is calculated based on the un - tagged dijet energy , the number of energetic charged particles not associated to either of the two @xmath23 jets , and @xmath97 , the larger of the two tau jet probabilities ( see figure [ fig : pdisc ] ) .
the event must satisfy @xmath980.9 .
events . ] after applying these cuts , the efficiency for signal @xmath22 decays is [email protected] , that for the @xmath32 background is @xmath99 , for @xmath30 and @xmath100 is @xmath101 while for @xmath21 events is @xmath102 .
the selection criteria yield 87 events of signal with 89 of background , corresponding to a relative statistical uncertainty of 0.15 on the determination of br(@xmath81 , @xmath103 ) .
the constraints on lcc-4 derived from this determination of the @xmath9 mass and width and other supersymmetric particle mass measurements at the lhc and a 1 tev linear collider , provide a prediction of the neutralino relic density with a relative accuracy of 0.18 , within the general mssm @xcite . the main contribution to
the remaining uncertainty comes from the weak constraint which data provide to mssm solutions where @xmath6 is significantly lower than its reference value for lcc-4 .
a detailed study shows that these solutions are all characterised by large values of the stau trilinear coupling , @xmath104 . in the mssm the @xmath105 coupling to the @xmath81 and @xmath9
bosons scales as @xmath106 and @xmath107 , respectively .
it has been proposed to determine @xmath104 through a measurement of the branching fraction of @xmath9 , @xmath108 @xcite . in the funnel region
the main neutralino annihilation mechanism is @xmath109 and @xmath110 .
the only @xmath9 decay into @xmath105s allowed by cp symmetry is @xmath111 which is kinematically forbidden for the lcc-4 parameters . and @xmath9 decay branching fractions as a function of the stau trilinear coupling @xmath104 as predicted by hdecay .
all the other mssm parameters have been kept fixed to those of the lcc-4 point . ]
however , at large values of @xmath112 , the @xmath113 decay gets a sizeable enhancement of its branching fraction . in this regime
, this channel also contributes to the neutralino annihilation rate through the @xmath114 process , thus lowering the corresponding relic density , as observed in the mssm scans . at the same time , a determination of the branching fraction of the decay @xmath113 , allows us to constrain the stau trilinear coupling .
figure [ fig : hdecay ] shows the decay branching fractions of the @xmath9 and @xmath81 bosons computed using the hdecay 2.0 program @xcite as a function of the @xmath104 parameter .
now , due to the same final state , a large @xmath115 yield can be detected by a standard @xmath116 analysis , such as that discussed in the previous section the present study shows that the branching fraction for @xmath81 , @xmath117 can be determined to @xmath4 0.15 and that for @xmath9 , @xmath27 to @xmath4 0.07 , from which a limit @xmath118 250 gev can be derived .
this constraint suppresses the tail at low values of @xmath6 bringing the prediction for the neutralino relic density to a relative accuracy of 0.08 , which is comparable to the current accuracy from the wmap data .
we have studied the @xmath0 process at @xmath1 = 1 tev using on fully simulated and reconstructed events for a supersymmetric benchmark point where the mass of the @xmath9 boson is 419 gev and the relic dark matter density in the universe crucially depends on its mass and width .
we find that the analysis of 2 ab@xmath35 of data should probide with relative accuracies of 1.0 gev and 2.9 gev in the heavy boson masses and widths , respectively .
the branching fractions of the @xmath119 decay can be determined with a 0.15 relative accuracy .
these data , in combination with other measurements available at the lhc and a @xmath5 linear collider , allows to infer the neutralino relic density in the universe with a relative accuracy of 0.08 .
we are grateful to abdel djouadi for pointing out the sensitivity of the @xmath120 decay branching fractions to the stau trilinear coupling and to michael peskin for discussion .
this work was supported by the director , office of science , of the u.s .
department of energy under contract no.de-ac02-05ch11231 and used resources of the national energy research scientific computing center , supported under contract no.de-ac03-76sf00098 .
e. a. baltz , m. battaglia , m. e. peskin and t. wizansky , phys .
d * 74 * 103521 ( 2006 ) [ arxiv : hep - ph/0602187 ] .
f. e. paige , s. d. protopopescu , h. baer and x. tata , arxiv : hep - ph/0312045 . g. belanger , f. boudjema , a. pukhov and a. semenov , comput . phys
. commun . * 176 * , 367 ( 2007 ) [ arxiv : hep - ph/0607059 ] .
m. battaglia , in the proc . of int .
conf . on linear colliders ( lcws 04 ) , paris , 2004 , vol . 2 , 951 [ arxiv : hep - ph/0410123 ] . s. agostinelli _ et al .
_ , nucl .
instrum .
a * 506 * 250 ( 2003 ) .
f. gaede , nucl .
. meth .
a * 559 * 177 ( 2006 ) .
t. behnke , pramana * 69 * , 697 ( 2007 ) .
f. gaede , t. behnke , n. graf and t. johnson , _ in the proceedings of 2003 conference for computing in high - energy and nuclear physics ( chep 03 ) , la jolla , california , 24 - 28 mar 2003 , pp tukt001 _ [ arxiv : physics/0306114 ]
. p. a. aarnio _ et al . _
[ delphi collaboration ] , nucl .
instrum .
a * 303 * 233 ( 1991 ) .
m. a. thomson , aip conf .
* 896 * 215 ( 2007 ) .
s. catani , y. l. dokshitzer , m. olsson , g. turnock and b. r. webber , phys .
b * 269 * , 432 ( 1991 ) .
d. j. jackson , nucl .
instrum .
a * 388 * 247 ( 1997 ) .
p. abreu _ et al .
_ [ delphi collaboration ] , eur .
j. c * 2 * , 581 ( 1998 ) .
t. sjostrand , p. eden , c. friberg , l. lonnblad , g. miu , s. mrenna and e. norrbin , comput .
commun . * 135 * , 238 ( 2001 ) [ arxiv : hep - ph/0010017 ] .
t. ohl , comput .
commun .
* 101 * , 269 ( 1997 ) [ arxiv : hep - ph/9607454 ] .
e. boos _ et al .
_ [ comphep collaboration ] , nucl .
instrum .
a * 534 * , 250 ( 2004 ) [ arxiv : hep - ph/0403113 ] .
s. y. choi , h. u. martyn and p. m. zerwas , eur .
j. c * 44 * , 175 ( 2005 ) [ arxiv : hep - ph/0508021 ] .
a. djouadi , j. kalinowski and m. spira , comput .
commun .
* 108 * 56 ( 1998 ) [ arxiv : hep - ph/9704448 ] . | this paper reports the results of a study of the @xmath0 process at @xmath1 = 1 tev performed on fully simulated and reconstructed events .
the estimated accuracies on the heavy higgs boson masses , widths and decay branching fractions are discussed in relation to the study of supersymmetric dark matter . |
dwarf galaxies are believed to represent the dominant population , by number , of the present day universe , and a major constituent of groups ( ct et al .
1997 ) and clusters ( e.g. ferguson & sandage 1991 ) .
studying their star formation and chemical enrichment histories is key to understanding the evolution of galaxies on cosmological timescales ( e.g. madau et al .
local group galaxies are ideal laboratories for detailed studies of dwarf galaxy properties : we can resolve their individual stars and thus learn about their star formation histories ( sfhs ) by exploring ages , metallicities , and the spatial distribution of the stellar populations they contain .
the color - magnitude diagram ( cmd ) is the best tool to retrieve the sfh of a stellar system .
the ideal situation is when it reaches the oldest main - sequence ( ms ) turnoffs with good photometric accuracy . in this case , the information on the sfh can be obtained directly and with little ambiguity from the distribution of stars on the ms , and its comparison with that predicted by stellar evolution models in this relatively well known phase of stellar evolution . in particular , the range of ages and metallicities present can be determined through comparison with theoretical isochrones . to quantitatively determine the sfh , it is necessary to compare the observed density distribution of stars with that predicted by stellar evolution models ( see gallart , zoccali & aparicio 2005 ) . shallower cmds ( e.g. those reaching just below the horizontal - branch ( hb ) or a couple of magnitudes below the tip of the red giant branch ( rgb ) ) still contain stars born through the whole galaxy s lifetime , but the interpretation of the distribution of stars in terms of the sfh is progressively less detailed and uncertain as the cmd is shallower .
the reason for this is twofold : on one hand , the stellar evolution models are less accurate for more advanced stellar evolution phases such as the rgb or the hb , because the corresponding physics is more complicated or uncertain . on the other hand , in these stellar evolution phases ,
stars of very different ages are packed together in the cmd in a small interval of color and/or magnitude , and suffer from important age - metallicity degeneracies ( see gallart 2000 and gallart et al .
2005 for detailed discussions of all these issues ) .
for all the milky way satellites , it is possible to obtain cmds reaching the oldest ms turnoffs using ground based telescopes , while for the rest of the local group , this is possible using the acs on board the hst .
the magellanic clouds ( mcs ) are particularly interesting among the milky way satellites since they are actively star forming galaxies which sfh can shed light on the role played by interactions in galaxy evolution .
but in spite of their proximity and intrinsic interest , it is remarkable that there are still important gaps in the knowledge of the sfh of the mcs .
this may be explained by their huge projected size and the big number of stars to analyze . in this paper
we will focus on the small magellanic cloud ( smc ) .
the smc is a dwarf irregular satellite of our galaxy , with low mean metallicity and a high mass fraction remaining in gaseous form ( van den bergh 1999 ) .
these characteristics suggest that the smc is in a more primitive evolutionary stage than its larger neighbor , the large magellanic cloud ( lmc ) .
historically , the study of this galaxy has been in general neglected in favor of the lmc .
the uncertainty about its line of sight depth , its more complicated shape ( elongated towards the northeast ) , and its larger distance from us are factors at play . to our knowledge ,
only two papers have presented cmds reaching the oldest ms turnoffs for a small field of view each : dolphin et al .
( 2001 ) and mccumber et al .
dolphin et al .
presented a combination of hst and ground - based _
v _ and _ i _ images of a smc field situated 2@xmath1 northeast of ngc 121 ( avoiding the contamination from 47 tucanae ) in order to derive the sfh of this smc field . from their ground - based images , they inferred a peak of star formation between 5 and 8 gyr ago with a medium age of 7.5 gyr at the 1 @xmath3 level , and that 14@xmath45% of the stars were formed more than 11 gyr ago .
the sfh they derived is also consistent at the 2 @xmath3 level with a continuous star formation from ancient times until @xmath02 gyr ago , when the star formation rate dropped to finally stop entirely in the past 0.5 gyr .
more recently , mccumber et al .
( 2005 ) analyzed the stellar populations of a smc field located in the `` wing '' area with observations from the hst wfpc2 .
they compared the luminosity function from their observed cmd with those obtained from two different model cmds , one with constant sfr(t ) and another with bursts of star formation at @xmath02 and at @xmath08 gyr .
they found that the population appears to have formed largely in a quasi - continuous mode , with a main period of star formation between 4 - 12 gyr ago and very recent star formation with bright stars as young as 100@xmath410 myr .
two main wide field studies of the smc exist : the pioneer work by gardiner & hatzidimitriou ( 1992 ) and a recent study from harris & zaritsky ( 2004 ; hereafter hz04 ) .
gardiner & hatzidimitriou ( 1992 ) presented the first large area study of the smc , and mainly concentrated their analysis in the outer regions , beyond 2@xmath1 from the smc center .
they took six photographic plates with the uk schmidt telescope covering a total area of 130@xmath1@xmath5130@xmath1 .
since their photometry is rather shallow ( reaching the hb level at r@xmath020 mag ) , they mainly gave information about the young populations ( age @xmath62 gyr ) .
from their cmds and contour plots of the surface distribution of ms stars with _ b - [email protected] and _ r_@xmath720 , they noticed the almost complete absence of bright ms stars in the northwestern part , while a considerable bright ms population was present in the eastern and southern area . with the aid of luminosity functions they found that young populations ( @xmath70.6 gyr in age ) are concentrated towards the center of the smc and in the `` wing '' region . using an index defined as the difference between the median color ( in _ ( b - r ) _ ) of the rc and the color of the rgb at the level of the hb
, the authors inferred that the bulk of the field population has a median age around 10 - 12 gyr .
they were the first to notice the different distribution of young and old populations in the smc .
zaritsky et al .
( 2000 ) and cioni et al .
( 2000 ) confirmed that the asymmetric appearance of the smc , similar to that of the hi ( stanimirovi et al .
1999 ) , is primarily caused by the distribution of young stars ( upper ms stars , younger than @xmath00.2 gyr ) , and that the older stars have a spheroidal distribution .
hz04 presented a study of the smc based on the magellanic clouds photometric survey ( mcps ) _ ubvi _ catalog ( zaritsky et al .
1997 ) , which covers a 4@[email protected]@xmath1 area of the smc to a depth of _ v_@xmath621 .
this work represents the most complete analysis of the spatially resolved sfh of the central part of the smc .
they divided the smc survey in a grid of 351 cells and obtained the sfh , through a @xmath8@xmath9 minimization between the star counts in the observed cmd and those in the model cmds based on isochrones by girardi et al .
they inferred that a significant fraction ( @xmath050% ) of all stars in the smc were formed more than 8.4 gyr ago . between 3 and 8.4 gyr
ago they found a quiescent period during which the smc formed few stars .
they also found a recent active time from 3 gyr ago until now , with bursts at 2.5 gyr , 400 myr , and 60 myr .
although deeper than the one from gardiner & hatzidimitriou ( 1992 ) , this study is still limited by its shallowness .
in fact , the oldest stars that they can observe on the ms are @xmath02 gyr old , and this implies a necessarily poor temporal resolution and larger uncertainties in their sfh at ages older than that . in spite of the increasing interest in studying the smc , reflected by the recent works mentioned before
, there are still many questions unanswered which require the information provided by cmds reaching the oldest ms turnoffs .
what is the age distribution of the old and intermediate - age population ?
, are there gradients in the composition of this underlying population ? .
shallower studies inform about the existence of a large amount of young population in certain areas , but does this young population reflect an exceptional increase of the star formation at the present time with respect to the average sfr ? . to shed light on these aspects , in this article we present twelve unprecedented deep _ br_-based smc cmds with positions ranging from @xmath01 to @xmath04 from the smc center .
our fields are distributed in different parts of the smc , avoiding the central area , where crowding may not allow us to reach the oldest ms turnoffs from the ground .
three of our fields are located in the `` wing '' area , near the field studied by mccumber et al .
( 2005 ) ; two fields are in the western part and one in the northwestern area , near the field from dolphin et al .
( 2001 ) located 2 northeast of ngc121 ; finally , six fields are placed towards the south at different galactocentric distances .
this strategic selection of fields , in spite of their small size , permit us to sample several regions of the smc with high temporal resolution in order to address the above issues .
the depth of our fields , reaching the oldest ms turnoffs with very good photometric accuracy , will allow us to obtain detailed sfhs from all of them , and to investigate , therefore , the variation of the sfh across the smc field .
this will be the subject of a forthcoming paper . here
we will focus the analysis in the inspection of the cmds with the aid of theoretical isochrones and color functions .
this paper is organized as follows . in
[ obs ] we present the observations and data reduction . in [ photometry ]
we discuss the photometry of the smc fields , obtained using daophot ii / allstar and allframe programs . in [ stellar ]
the cmds are discussed with the help of isochrones and color functions . in [ summary ] and [ conclu ]
our results are summarized and discussed . in a future paper ( nol et al .
, in preparation ) we will derive the sfh of the smc through comparison of the observed cmds with synthetic cmds .
the present paper is a result of a more comprehensive program to study the smc , which also includes the determination of its absolute proper motion by observing smc fields in which background quasars have been identified .
these quasars provide a quasi - inertial reference frame with respect to which ( in conjunction with existing radial velocity measurements ) the absolute total space velocity vector of the smc can be precisely determined ( see e.g. pedreros , costa & mndez 2006 ) .
the imaging strategy was therefore designed in a way that would satisfy the needs of the stellar populations program as well as the astrometry program .
given that for the latter it was mandatory to obtain _
r_-band images , this bandpass together with _
b _ imaging was chosen to construct the cmds .
the observations were made with a 24 @xmath10m / pixel tektronic 2048x2048 ccd detector attached to the cassegrain focus of the 100-inch irne du pont telescope ( c100 ) at las campanas observatory , in chile .
this combination gives a field size of 8.85@[email protected]@xmath11 and a scale of 0.259@xmath12/pixel .
inverse gain was 3e@xmath13/adu with a read noise of 7e@xmath13 .
basic reduction of the ccd frames was done using standard iraf tasks . for this purpose ,
bias exposures , sky flats , and dome flats were taken every night . throughout our four year campaign ( 2001 - 2004 ) , _
b _ and _ r _ band images of thirteen fields in the smc were obtained .
the coordinates of these fields and the data obtained for each of them are detailed in table [ fields ] , where the first column denotes the field , the second and third columns the right ascension and the declination respectively , fourth and fifth columns the galactocentric distance ( r ) and the position angle ( p ) , and the sixth and seventh columns the integration times in _ b _ and _ r_. seeing was typically between 0.@xmath127 and 1.@xmath122 in all epochs .
figure [ smchi]a shows a hi column - density image of the smc taken from stanimirovi et al .
( 1999 ) with our smc fields ( black squares ) and the fields from dolphin et al .
( 2001 ) and mccumber et al .
( 2005 ) ( black circles ) overlapped .
the black rectangle denotes the area covered by the mcps ( zaristky et al .
1997 ) . the actual figure from stanimirovi et al .
( 1999 ) includes the range in declination from @xmath0 - 75 to @xmath0 - 70 .
those squares denoted with `` qj '' represent the fields centered in quasars observed for astrometric purposes .
we followed the naming convention by tinney et al .
( 1997 ) , in which the numbers following `` qj '' refer to the right ascension of the quasar ( the declination , includded in the full tinney naming convention , was ommitted ) .
the fields denoted with `` smc '' ( and followed by the right ascension , for consistency with the naming convention ) were specifically selected for this project and were chosen to span a wide range of galactocentric distances , from @xmath01@xmath1 to @xmath04@xmath1 .
the combination of `` qj '' and `` smc '' fields results in a sampling of the smc that covers both a good range of galactocentric distances and position angles .
it is important to note that given the elongation of the smc , with a position angle of @xmath045 ( cioni et al .
2000 ) , a certain galactocentric distance in the northwest or southeast directions is different than in the southwest or northeast directions in which the stellar densities are larger . figure [ smchi]b show the distribution of young ms stars and figure [ smchi]c the layout of the rgb ( intermediate - age and old population ) stars in the smc from cioni et al . ( 2000 ) with our smc fields superimposed .
the name of each field was omitted for clarity .
rrrrrrrrr smc0057 & 00:57 & -73:53 & 65.7 & 164.4 & 1@xmath560 + 1@xmath5600 + 2@xmath5800 + 2@xmath51000 + 1@xmath51200 & 2@xmath560 + 2@xmath5600 + 3@xmath5800 + 1@xmath51200 + qj0037 & 00:37 & -72:18 & 78.5 & 294 & 1@xmath560 + 5@xmath5800 &
7@xmath560 + 16@xmath5600 + qj0036 & 00:36 & -72:25 & 79.8 & 288 & 2@xmath560 + 10@xmath5600 + 12@xmath5800 & 15@xmath560 + 26@xmath5500 + 22@xmath5600 + qj0111 & 01:11 & -72:49 & 80.9 & 89.5 & 1@xmath560 + 3@xmath5800 & 8@xmath560 + 4@xmath5500 + 17@xmath5600 + 1@xmath5700 + 1@xmath5800 + qj0112 & 01:12 & -72:36 & 87.4 & 81 & 1@xmath560 + 1@xmath5600 + 6@xmath5800 & 9@xmath560 + 7@xmath5500 + 15@xmath5600 + qj0035 & 00:35 & -72:01 & 95.5 & 300.6 & 4@xmath5600 + 1@xmath5700 & 4@xmath560 + 1@xmath5100 +
3@xmath5500 + 15@xmath5600 + 3@xmath5800 + qj0116 & 01:16 & -72:59 & 102.5 & 95.2 & 1@xmath560 + 13@xmath5600 + 2@xmath5800 & 7@xmath560 + 14@xmath5500 + 20@xmath5600 + smc0100 & 01:00 & -74:57 & 130.4 & 167.5 & 1@xmath560 + 7@xmath5800 + 3@xmath5900 & 1@xmath560 + 7@xmath5600 + 3@xmath5700 + qj0047 & 00:47 & -75:30 & 161.7 & 187.7 & 1@xmath5800 + 3@xmath51000 & 6@xmath560 + 16@xmath5600 + 2@xmath5800 + qj0033 & 00:33 & -70:28 & 172.9 & 325 & 1@xmath560 + 5@xmath5600 + 1@xmath5700 & 6@xmath560 + 3@xmath5500 + 20@xmath5600 + smc0049 & 00:49 & -75:44 & 174.8 & 184.6 & 1@xmath560 + 1@xmath5600 + 4@xmath5800 & 1@xmath560 + 5@xmath5600 + qj0102 & 01:02 & -75:46 & 179.5 & 169.4 & 3@xmath560 + 6@xmath5800 & 8@xmath560 + 3@xmath5500 + 15@xmath5600 + 1@xmath5700 + 2@xmath5800 + smc0053 & 00:53 & -76:46 & 236.3 & 179.4 & 1@xmath560 + 8@xmath5800 + 2@xmath5900 & 1@xmath560 + 8@xmath5600 + 2@xmath5700 + [ fields ]
we performed our photometry using the daophot ii / allstar software package ( stetson 1987 , and updates : stetson 1990 , 1992 ) to find stars and to determine an empirical psf for each frame and then simultaneously fit it to all detected stars . after testing various psf models , we adopted a moffat function with @xmath14=2.5 ( stetson et al .
2003 ) as the best analytical model .
the moffat function is usually a much better representation of a star image than a gaussian function , because it has more prominent and extended wings , like real star images .
also , unlike other functions ( e.g. the lorentz function ) , the moffat function offers the option of tuning the radial falloff of the wings to match observations . between 50 and 100 stars
were used to construct the psf of each image .
quadratic spatial variation was included because it reduced the fitting errors .
after the daophot / allstar analysis of all the original frames was performed , an optimum , complete star list was achieved by cross matching the allstar results of the individual frames with daomaster ( stetson 1993 ) .
allframe ( stetson 1994 ) was then used to obtain simultaneous photometry of stars in all ccd images of each smc field .
after allframe , we ran montage2 for each field
. montage2 takes the images from which all the stars have been subtracted according to their final positions and brightnesses , and applies the known geometric transformations that relate their coordinate systems and the known differences in their photometric zero - points , to produce a median image of each field .
we tested in each of the 13 smc fields that the number of stars detected on the median of the subtracted images , and considered real stars after running daophot , allstar , and allframe a second time , was less than 1% , i.e. negligible for our goal .
allframe produces an image - quality index , chi , which is a dimensionless measure of the agreement between the brightness profile of any given object and that of the model psf for the frame in which it is measured .
similarly , allframe gives the sharp index , which is a first - order estimate of the intrinsic angular radius of a source .
another index provided by allframe is @xmath3 , which represent the standard error of the star s magnitude and is representative of the internal errors of the photometry .
stars with at least one valid measurement in each band ( _ b _ and _ r _ ) were selected , and their final magnitudes were obtained using daomaster , which combines the magnitudes measured for each star in each image to provide the `` mean weighted '' magnitude .
daomaster also provides @xmath3 , chi , and sharp parameters for each stars , which are a combination of the corresponding parameters for each image .
we used the following limits for the error and shape parameters given by daomaster : @xmath3@xmath15@xmath16=@xmath3@xmath17@xmath16+@xmath3@xmath18@[email protected] , -0.6@xmath19@xmath20sharp@[email protected] , and @xmath20chi@[email protected] .
the final number of stars we kept , measured in _ ( b - r ) _ of each field ( and consequently , in each cmd ) is given in table [ exposure ] . a total of 215,121 stars down to _
r_@xmath2124 were measured .
ccccccccccccccccc smc0057 & 21246 + qj0037 & 21579 + qj0036 & 23756 + qj0111 & 25845 + qj0112 & 47548 + qj0035 & 24114 + qj0116 & 19875 + smc0100 & 7801 + qj0047 & 9762 + qj0033 & 3797 + smc0049 & 3580 + qj0102 & 4030 + smc0053 & 2188 + [ exposure ] the so - called aperture corrections , i.e. , the corrections that place the relative profile - fitting magnitudes on the system of the `` total '' instrumental magnitudes for a particular frame , were obtained from synthetic aperture photometry by measuring several isolated , bright stars through a series of increasing apertures and the construction of growth curves ( stetson 1990 ) .
we used a growth sequence which consists in twelve apertures from _
@xmath22 _ to _ @xmath23 _ in such a way that we considered the first radius _
@xmath22 _ about half of the one we used for the aperture photometry of our objects and _ @xmath23_=30@xmath12 .
the sequence is _
@xmath24=@xmath25{r_{12}/r_{1}}$]@xmath5 @xmath26 _ , _ @xmath27_. the aperture corrections were derived using the program daogrow ( stetson 1990 ) . in each image we selected the brightest and most isolated 30 to 80 stars among those used in the derivation of the psf for that frame .
the aperture photometry results were then provided to daogrow which returns the `` total '' instrumental magnitude and its standard error for each of the selected stars . the aperture correction for a particular frame
was obtained as the median of the differences between the `` total '' magnitude and the profile - fitting allframe magnitude for all selected stars on that frame .
errors in the aperture corrections were calculated as the standard error of the mean of these differences , and were typically between @xmath40.001 and @xmath40.003 .
our instrumental photometric system was defined by the use of the harris _ bvri _ filter set , which constitutes the default option on the c100 for broad - band photometry on the standard johnson - kron - cousins system .
in photometric nights , typically six _ bvri _ standard star areas from the catalog of landolt ( 1992 ) were observed several times to determine the transformation of our instrumental magnitudes to the standard _
br _ system .
thirty - one different standard stars with colors : -0.49@xmath7_(b - r)[email protected] were observed , and a total of 272 observations were made ; 178 in 2001 and 94 in 2002 .
most of these areas include stars of a wide variety of colors .
a few of them were followed up to about 2.0 airmasses to determine atmospheric extinction optimally . during these nights
we also obtained short exposure _ br _ frames of all our fields of interest , which served the purpose of calibrating all frames taken with non - photometric sky .
field qj0035 is not considered in this paper because we do not have yet a secure calibration for it .
the landolt standard images are uncrowded fields , thus no profile - fitting photometry was necessary for them . instead , daogrow was used in an identical way as for the program stars to directly derive the total instrumental magnitudes for the standard stars from their aperture photometry . to put our observations into the standard system
, we used the following transformation equations : @xmath28 where @xmath29 and @xmath30 are the instrumental and standard magnitudes respectively , and ( x@xmath31 , x@xmath32 ) are the airmasses . no time dependence terms were added since a preliminary fit showed no trends in the residuals of both _ b _ and _ r _ with time .
the color terms ( @xmath14@xmath31 , @xmath14@xmath32 ) and extinction coefficients ( @xmath33@xmath31 , @xmath33@xmath32 ) , as well as the zero - points ( @xmath34@xmath31 , @xmath34@xmath32 ) are unknown , presumably constant , transformation coefficients and are to be calculated . both the color - dependent term and the zero - point in the transformation ,
are expected to be reasonably constant properties of the telescope / filter / detector combination and , in fact , the night - to - night differences for a given year in their computed values were found to be consistent with the uncertainty of the individual determinations .
the above equations were applied to the landolt standard star magnitudes , and solved for @xmath34@xmath31 , @xmath34@xmath32 , @xmath14@xmath31 , @xmath14@xmath32 , @xmath33@xmath31 and @xmath33@xmath32 , for each night , using a custom program .
ccrrrrrrrrcrl 2001 & [email protected] & [email protected] & + [email protected] & + [email protected] + 2002 & [email protected] & [email protected] & + [email protected] & + [email protected] + [ coefficients ] then , we followed an iterative procedure to refine our photometric transformation .
first , a new set of unique ( @xmath34@xmath31 , @xmath34@xmath32 ) and ( @xmath14@xmath31 , @xmath14@xmath32 ) values for each campaign was obtained by imposing the extinction coefficients ( @xmath33@xmath31 , @xmath33@xmath32 ) corresponding to each night .
then , we applied the resultant zero - points and color terms to each night of each year and new extinction coefficients were derived for each night . in this way
, we have a set of ( @xmath34@xmath31 , @xmath34@xmath32 ) and ( @xmath14@xmath31 , @xmath14@xmath32 ) values for each year and ( @xmath33@xmath31 , @xmath33@xmath32 ) values for each night of each year . in table
[ coefficients ] we present the zero - points and the color coefficients .
the first column denotes the year , the second and third columns indicate the ( @xmath34@xmath31 , @xmath34@xmath32 ) coefficients , and fourth and fifth columns the ( @xmath14@xmath31 , @xmath14@xmath32 ) coefficients .
the final values for the extinction coefficients are shown in the second and third columns of table [ atmospheric_extinction ] , where the first column shows the night to which the coefficients correspond .
figure [ four ] shows the standard distributions for each filter and each year ; the slope represents the extinction coefficient .
fitting errors in the zero - points correspond to @xmath3/@xmath35 , where n is the number of measurements we have .
they turned out to be between @xmath40.001 and @xmath40.002 .
the total zero - point errors of the photometry , including the error in the extinction , in the aperture corrections , and the uncertainties in the calibrations , are @xmath00.02 mag in both _ b _ and _ r_. these values are consistent with the systematic errors derived comparing photometry of different epochs .
using those smc fields for which we have standard star observations from more than one night , we tested the photometric differences resulting from the use of different standard sets corresponding to different nights .
we found that these differences are between @xmath40.001 and @xmath40.03 in _ b _ and between @xmath40.001 and @xmath40.04 in _ r _ ( standard errors of the mean ) .
ccc 16/10/2001 & [email protected] & [email protected] + 17/10/2001 & [email protected] & [email protected] + 18/10/2001 & [email protected] & [email protected] + 19/10/2001 & [email protected] & [email protected] + 10/10/2002 & [email protected] & [email protected] + 13/10/2002 & [email protected] & [email protected] + [ atmospheric_extinction ]
we present unprecedented deep smc cmds obtained using a medium aperture telescope ( 100-inch ) .
the quality of the cmds is comparable to that of the cmds obtained for the mcs using the hst ( e.g. mccumber et al . 2005 ) . given the depth of our diagrams , which reach the oldest turnoffs , accurate information about age can be derived from the ms turnoff luminosities . as discussed by gallart , zoccali , & aparicio ( 2005 ) , reaching the oldest turnoffs will allow us to break the age - metallicity degeneracy affecting most methods to obtain the sfh .
figures [ three_east ] to [ six_south ] show the [ _ ( b - r),r _ ] cmds for the twelve smc fields in order of increasing galactocentric distance .
the fields range from galactocentric radius @xmath01@xmath36 to @xmath04@xmath36 , and sample different regions of the smc , as shown in figure [ smchi ] , where the fields are depicted in the context of the hi distribution ( figure [ smchi]a ) , and of the young ( figure [ smchi]b ) and intermediate - age and old ( rgb stars ) populations ( figure [ smchi]c ) .
all the cmds reach a couple of magnitudes below the old ms turnoff ( even the most populated ) , with great photometric accuracy .
figure [ three_east ] shows the cmd of fields qj0111 , qj0112 , and qj0116 , which have a galactocentric radius ranging from 1.4@xmath36 to 1.7@xmath36 .
these fields correspond to the eastern side of the galaxy ( facing the lmc ) as seen in figure [ smchi]a , from which we can notice that they are located in an area of high hi concentration , inside the supergiant shell 304a with hi mass of 5.7@xmath510@xmath37 m@xmath38 ( the total hi mass in the area observed by stanimirovi et al .
is @xmath04@xmath510@xmath39 m@xmath38 ) .
these fields are also in the region of the bridge which connects the so - called `` stellar bar '' with the `` wing '' . from figure [ smchi]b and figure [ smchi]c
, it can be noticed that these fields are located in very densely populated isopleths of young , intermediate - age and old population .
indeed , the cmds of these fields show a conspicuous ms , well populated from the oldest turnoff at r@xmath022 to r@xmath016 , which implies the presence of a large number of very young stars in all the cmds .
the three western fields are also shown in figure [ smchi]a and the corresponding cmds are presented in order of increasing galactocentric distance from 1.3@xmath36 to 2.9@xmath36 in figure [ three_west ] .
these fields are situated in the opposite side of the lmc , in a region with low concentration of hi . from figure
[ smchi]b it is evident that they are located in an area of low young stellar density , but still high density of intermediate and old population similar to that of the eastern fields .
the cmds of these western fields show a less significant ms , even at similar distances from the center as the eastern fields .
this fact is in agreement with the low density of the hi column in this part of the galaxy ( see figure [ smchi]a ) .
we would like to address if the differences seen in the young population can be extended to the intermediate and old population .
figure [ six_south ] shows the cmds of the six southern fields , whose galactocentric distances range from 1.1@xmath36 to 4@xmath36 .
as seen in figure [ smchi]a these fields are located in regions in which the hi column - density is very low .
field smc0057 , the closest to the center , presents a conspicuous young ms . in figure
[ smchi]b is shown that the stellar density of young stars in this field is higher than in the rest of the southern fields and similar to the eastern field qj0111 .
field smc0100 still shows many bright young ms stars but further on 2.7@xmath36 the cmds have less populated young ms , and present a similar distribution of stars ( with less statistics while going further south ) .
the four furthest fields are located in an area in which the stellar density is similar , as shown in figures [ smchi]b and [ smchi]c . depending on their mass and metallicity , the core he - burning stars produce the horizontal branch ( hb ) and the rc .
the presence of a prominent blue hb points out the existence of an old , metal poor population ; the lack of such hb and the presence of a rc may indicate that core he burners are not so old , or more metal rich , or both ( see gallart et al .
2005 for details ) .
a shared feature in all our smc cmds is the absence of a well populated , blue extended hb , pointing out that the amount of field stellar population as old and metal poor as that of the milky way halo globular clusters and dwarf spheroidal galaxies is small in the smc .
the rc is well populated and extended in luminosity , with a width of up to @xmath40r@xmath01 mag , denoting the presence of a large intermediate - age population . the fields with a prominent bright ms ,
as the eastern fields and smc0057 , for example , show the _ vertical extension of the rc _ feature corresponding to non - degenerate core he - burners ( gallart 1998 ) . the older stars in the core he - burning phase lie in the lower part of the observed rc , while younger stars are brighter .
in the following sections , we discuss the main features of the smc cmds using isochrones ( sec [ isochrones ] ) and color functions ( sec [ cf ] ) .
this will allow us to describe the smc stellar populations , as a starting point to a study ( nol et al .
, in preparation ) in which we will address the smc sfh by comparing in detail the distribution of stars in the observed cmds with a set of model cmds .
the interpretation of cmds of composite stellar population strongly relies on the stellar evolution models adopted ( see gallart et al . 2005 ) . for the present work
, we used the teramo stellar evolution models ( pietrinferni et al .
2004 ) as the reference . in figures
[ three_east_iso ] , [ three_west_iso ] , and [ six_south_iso ] we have superimposed teramo isochrones to our cmds for three different metallicities suitable for the smc stellar populations : z=0.001 ( [ fe / h]=-1.27 ) , z=0.002 ( [ fe / h]=-0.96 ) , and z=0.004 ( [ fe / h]=-0.66 ) . the final selection of the metallicities was done taking into account the rgb and ms star colors and its best match by teramo isochrones of given ages and metallicities .
the observed cmds have been transformed to absolute magnitudes using a distance modulus of @xmath41 ( see van den bergh 1999 ) and de - reddened as follows .
for the two outermost fields , the reddening values given by the iras / cobe extinction maps ( schlegel , finkbeiner & davis 1998 ) were used .
for the innermost regions , schlegel et al .
( 1998 ) estimate a typical reddening of e(b - v)=0.037 , from the median dust emission in surrounding annuli , since reddenings _ through _ the smc can not be calculated because its temperature structure is not sufficiently well resolved by dirbe .
their quoted value , therefore , is not accurate enough and we have estimated a mean value of the reddening for each of the fields smc0057 , qj0037 , qj0036 , and qj0111 by requiring a good fit of the cmd by the same isochrones that were overlapped to the other fields , for which the iras / cobe reddening values are reliable .
this assumes the same age - metallicity relation for all fields , which is in agreement with the results found by carrera ( 2006 ) and carrera et al .
( 2006 , in preparation ) using caii triplet spectroscopy . in all cases , the relations @xmath42e(b - v ) , and @xmath43 e(b - v )
( schlegel et al .
1998 ) have been assumed to calculate the extinction in the _ b _ and _ r _ bands .
canonical isochrones of 6 , 9 , and 13 gyr with z=0.001 , isochrones with overshooting of 1 , 3 , and 4 gyr with z=0.002 , and of 0.1 gyr with z=0.004 were overlapped . in the eastern fields and in field smc0057 a 0.03 gyr isochrone with overshooting and z=0.004 was also superimposed .
figure [ three_east_iso ] shows the cmds of the eastern fields with the superposition of teramo isochrones . in these cmds
the conspicuous ms is well populated from the oldest turnoff at m@xmath18=3.5 ( 13 gyr isochrone ) up to the 0.03 gyr isochrone .
all the cmds show a large number of young stars well represented by the 0.03 gyr and 0.1 gyr isochrones .
the areas around the 1 to 6 gyr isochrones are densely populated , indicating a strong presence of intermediate - age stars .
no obvious differences between the ms of these fields can be inferred from the comparison with isochrones alone .
figure [ three_west_iso ] shows the cmds of the western fields with the isochrones overlapped . in these fields , the intermediate - age population ( @xmath03 - 4 to 9 gyr ) seems dominant .
fields qj0036 and qj0037 have a small fraction of stars born up to 0.1 gyr ago , while qj0033 , the most remote of the three , seems to have no stars younger than 1 gyr ( though this could be an effect of small number statistics ) .
figure [ six_south_iso ] shows the cmds of the six southern fields with the superposition of isochrones .
field smc0057 , the closest to the center , presents an important ms where the 0.1 gyr isochrone is still quite populated . in field
smc0100 there is still a substantial population between 1 and 3 gyr old , but fields qj0047 to smc0053 show few stars younger than 3 - 4 gyr .
the population of these relatively far - off fields is not purely old but contains an important amount of intermediate - age stars that inhabit the zones of the ms and subgiant branch around the 3 , 4 , 6 , and 9 gyr isochrones , in addition to the several old stars around the 13 gyr isochrones .
gallart et al .
( 2005 ) have discussed the potential of ms color functions ( cfs ) to provide information on the stellar populations present in a galaxy .
they concluded that , in fact , the cf may be a better tracer of the sfh than the lf .
the reason is that , in cmds like [ _ ( b - r ) , r _ ] or [ _ ( v - i ) , i _ ] , the ms of isochrones of different age follow roughly vertical paths up to the ms turnoff where they turn to the red , almost perpendicularly to the ms , to form the subgiant branch or , in the case of more massive stars , the blue loops .
this effect is clear from the cmd in panel ( a ) of figure [ docfsynt ] ( we will return to this figure later ) .
the fact that the turnoff of a stellar population gets redder as the populations gets older implies a general dependency on age of the cf of the ms and subgiant branch stars brighter than the oldest ms turnoff . here
we will discuss cfs that will include the whole cmd within a region comprised between : -1@xmath19m@[email protected] and -0.5@xmath19(b - r)@xmath191.5 .
the star counts have been normalized to the total stars present in the integrated range .
most of the information belongs in any case to the ms and subgiant branch .
the fraction of cf that belongs to the rgb and rc serves as a check of the metallicity assumed for the synthetic cmd , and contributes to the overall normalization of the cf .
comparison of the stellar content of a galaxy with isochrones and cfs is complementary in terms of the information on the stellar population they provide .
the isochrones give us the range of ages present in a given cmd ; the comparison between theoretical and observed cfs is a good tool to assess the relative amounts of stars of different ages present in a stellar population .
this is a particular case of a comparison of the number of stars in a set of boxes in the observed and synthetic cmds respectively ( e.g. gallart et al . 1999 ) .
however , the cf is particularly useful for a first qualitative assessment of the sfh because it is a one - dimensional representation of the content of the cmd in which there is a relatively direct correlation between range of ages and color .
all the models used here have been computed using the synthetic cmd algorithm iac - star ( aparicio & gallart 2004 ) .
panel ( a ) in figure [ docfsynt ] shows a synthetic cmd calculated assuming a constant sfr(t ) from 13 gyr ago to date and metallicities : z=0.004 in the range 0 gyr@xmath19t@xmath191 gyr , z=0.002 in the range 1 gyr@xmath19t@xmath195 gyr , and z=0.001 in the range 5 gyr@xmath19t@xmath1913 gyr ( see inset ) .
stars corresponding to each age interval are depicted in different colors .
a 30% of binary stars have been assumed ; the parallel feature to the red of the ms between 3.5@xmath19m@xmath18@xmath195 corresponds to binary stars .
some of these binaries between 8 - 13 gyr old are noticeably brighter than single stars at the same age and mass ( red points in the synthetic cmd between 2@xmath19m@xmath18@xmath193 ) .
panel ( b ) in figure [ docfsynt ] shows the composite cfs of populations with different sfr(t ) ( shown in the inset ) .
three main features characterize the cf : a local elevation in the blue part between -0.3@xmath19(b - r)@xmath190.2 , a central peak between 0.2@xmath19(b - r)@xmath190.9 , and a second local maximum in the red part .
panel ( c ) of figure [ docfsynt ] shows which populations contribute to each feature in the cf for the synthetic cmd in panel ( a ) . by comparing these , we find the following : the elevation in -0.3@xmath44(b - r)@xmath440.2 in the composite cf mostly corresponds to stars in the age range 0@[email protected] gyr ; in the central peak ( 0.3@xmath44(b - r)@xmath440.9 ) there are populations of all ages , predominating the intermediate - age and old stars ( @xmath02 - 13 gyr ) .
the second local maximum corresponds to the rc and the red part of the subgiant branch and is composed by stars with age @xmath451 gyr as seen in panel ( c ) in figure [ docfsynt ] .
going back to panel ( b ) , the cf in black corresponds to a galaxy with a constant sfr(t ) ( i.e. , the cf of the cmd in the panel ( a ) ) .
the blue cf is from a synthetic cmd computed assuming a sfr(t ) that increased from 3 gyr ago to the present .
this cf presents a higher blue elevation due to the bigger number of young blue stars ; also the central peak is somewhat blueshifted .
the green cf represents a stellar population in which the sfr(t ) dropped by 50% between 3 and 8 gyr ago .
this is reflected in the shape of the absolute maximum , where a depression appears as a result of the assumed sfh .
this kind of feature would be difficult to detect in a real population .
finally , the red cf corresponds to a sfr(t ) with a totally quiescent period between 3 and 8 gyr ago .
the deep hollow shown in the absolute maximum of this cf is due to the gap at intermediate ages in the sfh .
the last two cfs show a higher blue elevation as compared to a cf of a cmd with constant sfr(t ) due to the higher fraction of young stars over the whole population . prior to present a discussion based on cfs , we will briefly talk about the crowding effects and , in general , the observational errors that are present in our photometry .
a careful study of the extent and nature of these crowding effects is key to study the stellar populations present in the smc
. we will understand and quantify the crowding effects using the artificial - star tests procedure .
these artificial - star tests have been performed in a way similar to that described in gallart et al .
( 1999 ) and will be presented in detail in a forthcoming paper ( nol et al . in preparation ) . in short ,
a number of artificial stars of known magnitudes and colors are injected in the original frames , using the addstar algorithm of daophot ii ( stetson 1993 ) .
the photometry is re - done following the same procedure used to obtain the photometry for the original frames .
this process is repeated several times in order to test a large number of artificial stars ( 60,000 in the case of each of our smc fields ) . the injected and recovered magnitudes of the artificial stars , together with the information from those that were lost ,
provide the necessary information about all the observational effects present in the photometry .
we will discuss here how the observational errors affect the cf of a few representative fields .
those located closer to the smc center ( e.g. smc0057 and qj0111 ) are more affected by crowding than the rest of the fields . to quantify these observational errors
we have simulated them for different smc fields in the synthetic cmd of figure [ docfsynt ] .
following gallart et al .
( 1999 ) , we will call the latter `` model cmds '' .
the model cfs ( corresponding to the model cmds ) corresponding to fields smc0057 , qj0111 , and smc0049 are shown in figure [ dispersadas ] , together with the original cf ( the cf obtained from the synthetic cmd ) .
the observational errors may produce significant variations in the appearance of the cf of our more central , crowded field smc0057 . simulating the observational errors of this field in the synthetic cmd
, we obtain a model cf ( dotted line ) which differs substantially from the cf of the original synthetic cmd : the absolute maximum in its cf is smaller than in the cf of the synthetic cmd , and the blue elevation is bigger because faint stars on the low ms are lost much more frequently than the brighter blue stars that constitute the young ms .
the dispersion introduced in the cmd when the observational errors are simulated produces an absolute maximum that is slightly wider as shown in the dotted line in figure [ dispersadas ] .
this shows how crowding could simulate a non - constant sfh , mimicking a recent enhancement of the star formation , similar to the one shown in blue in panel ( b ) of figure [ docfsynt ] .
although still present , the effects of crowding are less severe in the case of field qj0111 since the cf of the corresponding model cmd ( dashed line ) resembles quite closely the cf for the synthetic cmd . in this field
, confusion is mainly affecting the _ yellow _ stars so the main maximum is shorter than in the synthetic cf . in the case of smc0049
, it can be seen that the cf of the model cmd ( dot - long dashed line ) is similar to the one of the synthetic cmd ( solid line ) , indicating that crowding is affecting very little the cmds of this part of the galaxy ( at @xmath02.9@xmath36 ) . in conclusion , only the cf of field smc0057
is substantially affected by observational errors .
the effects of crowding on field qj0111 can be considered as an upper limit for fields qj0112 and qj0116 . for fields
smc0100 , qj0047 , qj0102 , smc0053 , qj0037 , qj0036 , and qj0033 the crowding level is similar to that of smc0049 . therefore , we can safely discuss the stellar content of all of our fields ( except smc0057 ) by comparing their cfs straight with that of synthetic cmds .
let s concentrate now on the observed cfs of each of our fields . in figures [ three_east_synt ] to [ six_south_synt ]
the cf corresponding to the synthetic cmd ( dashed lines ) from panel ( a ) in figure [ docfsynt ] and the cfs ( solid lines ) from the observed cmds were overlapped . note that the shape of the synthetic cf depends on the selection of metallicities and stellar evolution libraries .
however , small changes in the metallicity law , which result in synthetic cmds with a color distribution compatible with the data have also very similar cfs . comparison between the observed and synthetic cfs computed assuming constant sfr(t ) lets us to qualitatively analyze the sfh of each field .
figure [ three_east_synt ] shows the cfs of the eastern smc fields . in the first panel , corresponding to field qj0111 , synthetic cfs of synthetic populations with bursts of 50% 1 gyr ago ( long dashed line ) , of 30% 1 gyr ago ( dotted line ) , and of 50% 2 gyr ago ( dot - long dashed line ) were also overplotted .
the cf of field qj0111 shows some differences with respect to the cf of a population with constant sfr(t ) . in particular , the observed cf has higher values than the synthetic one in the blue elevation ( -0.3@xmath44(b - r)@xmath440.2 ; [email protected] gyr ) which indicates a sfh with enhanced star formation at young ages .
this fact is also shown by comparing with the cf of synthetic populations with bursts .
in fact , the blue maximum and the main peak of the cf with a burst of 30% 1 gyr ago and the ones of the observed cf match very well , pointing out that there was an episode around 1 gyr ago of enhanced star formation in this field which may have continued to the present time . though a comprehensive analysis of the cmds will be presented in a forthcoming paper , this superposition allows us to say that the observed cf is compatible with a sfh with an increase in the star formation of @xmath030% in the last 1 gyr .
also note that crowding effects are not very significative in this field , as seen from the artificial star tests .
the cf of field qj0112 is a remarkable case where the cf corresponds almost exactly with that of a population with constant sfr(t ) ( at least in average ) .
the cf of qj0116 presents a somewhat greater values of the cf in 0@xmath44(b - r)@xmath440.3 as compared with the cf of a cmd with constant sfr(t ) , which may correspond to stars formed between @xmath01 - 2 gyr ago , possibly indicating some enhancement in the sfr(t ) at those ages in comparison with more recent ages . note that these three fields have a very similar cmd , and that simple comparison with isochrones gives the same information for all three : presence of stars of all ages .
the differences in their cfs are a good illustration of the complementary information that the cf can provide .
the presence in the observed cfs of all the features that characterize the synthetic cf is the counterpart , in this representation of the data , of the fact that the cmds are well populated over all age ranges . the additional information brought forward by the cfs is the relative weight of the different age ranges in the population . in figure [ three_west_synt ] , the cfs of the western smc fields are shown .
the almost complete absence of the blue elevation in the color interval -0.3@xmath44(b - r)@xmath440.2 , reflects the fact that there are few stars younger than @xmath01 gyr .
this is another example of the usefulness of the cfs : while in fields qj0036 and qj0037 there are stars until around the 0.1 gyr isochrone , the cfs indicate that the relative amount of stars between 0 gyr and 2 gyr is substantially less than in the case of a field with constant sfr(t ) .
the central maximum is slightly shifted to the red , possibly indicating that the sfr(t ) started to decline around 3 gyr ago .
note that this shift is unlikely due to an overall shift of the observed cmd with respect to the synthetic one , since the shift does not affect the red relative maximum of the cf .
in fact , this maximum is slightly shifted to the blue and higher in the cf of the observed cmd as compared with that of the synthetic one .
this is possibly due to field stars , which contribute more in relative numbers to the poorly populated cmds . actually , the observed red maximum is higher with respect to the theoretical one in the less populated qj0033 .
the composition of the population in these three western cmds seems to be very similar , even in the outermost field , qj0033 .
finally , figure [ six_south_synt ] shows the cfs of the southern smc fields . in the case of field smc0057
, the relative amount of young stars might be overestimated due to crowding effects if these were not taken into account .
the overplot of the observed cf ( solid line ) with the synthetic ( dashed line ) and the model ( i.e. after taking into account crowding effects , dotted line ) ones shown in figure [ six_south_synt ] indicate that part of the enhancement of the blue elevation in field smc0057 is produced by observational errors but that there is still a true recent enhancement ( 0@xmath19t@xmath193 gyr ) in the star formation .
the cf of field smc0100 is still almost compatible with a constant sfr(t ) on average , with possibly a truncation of the sfr(t ) at recent ages , as also hinted by the comparison of its cmd with isochrones in figure [ six_south_iso ] .
moving further away from the smc center , the elevations corresponding to the first structure in the cfs are less conspicuous , indicating that young stars are less common and intermediate - age stars dominate the fields .
the fact that the main peak is redder from field qj0047 on ( except , maybe , in the case of field qj0102 which is situated more towards the east than the other southern fields ) , points out that there is a larger amount of old stars and , like in the western fields , that possibly the sfr(t ) started to decline around 3 gyr ago . as in the case of the western field qj0033 , the second , red relative maximum in the cfs gets bigger with increasing galactocentric distance .
the reasons for this are twofold .
first , the amount of stars in the rc and in the rgb , relative to the total stellar content is larger at larger distances from the center ( even though not large differences among fields seem to be present beyong [email protected]@xmath36 ) , where the age of the stellar population is older on average .
second , the contamination by foreground stars is larger relative to the total number of stars from the galaxy .
this last factor , may be the main player in the case of field smc0053 .
we obtained _ b _ and _ r_-band photometry of stars from 12 smc fields observed during 4 different campaigns with the c100 telescope at las campanas observatory .
the spatial distribution of the fields samples different parts of the galaxy , both in areas with large amounts of recent star formation such as the `` wing '' area and `` undisturbed '' regions at the west and towards the south part of the galaxy , in which the stellar populations may be representative of the underlying population of the smc formed prior to the events of star formation that shaped its current irregular morphology .
the stellar content present in these smc fields has been analyzed by means of a set of [ _ b - r , r _ ] cmds that reach the oldest ms turnoffs ( m@[email protected] ) with an excellent photometric accuracy . in this first aproach to study the sfh of our smc fields , we found the following .
the fields in the `` wing '' area contain very young blue stars and show a broad ms turnoff / subgiant region and a wide range in luminosity of the rc , pointing out that star formation in these parts has extended from at least @xmath013 gyr ago to the present with no substantial gaps , as the areas around all the overlapped isochrones are well populated .
( the closest to the center of the smc in this `` wing '' area ) seems to have experienced an enhancement in its recent star formation , compatible with an enhancement of 30% in the sfr(t ) over the last 1 gyr .
this is indicated by the height of the blue elevation its cf ( color interval in the range -0.3@xmath44(b - r)@xmath440.2 ) as compared with that of cfs corresponding to synthetic populations with different sfr(t ) .
the other two fields analyzed in this direction seem to have had a sfr(t ) close to constant on average .
very little star formation has been going on from @xmath01 gyr ago until now in the western fields , as indicated by the much lower amount of young blue stars and the lack of the blue elevation in the cfs .
considering the rc ( formed by stars older than @xmath01 gyr ) of fields qj0036 and qj0111 , both located at the same galactocentric distance , we found that the number of stars is comparable , indicating that the integrated sfr(t ) from old ages and until @xmath01 gyr ago in both fields is similar , and that there was a mechanism in the `` wing '' area that triggered the star formation around 1 gyr ago , in particular , in the location of field qj0111 . field qj0033
, though located further away from the center in the northwest , presents a distribution of stellar ages similar to that of the two other western fields , as seen from the resemblance of the three cfs and from its cmd , in which the area below the turnoff of the 4 gyr isochrone is still quite populated .
in fact , the apparent differences between the cmd of field qj0033 and the other two are due to the much larger number of stars populating those .
if only a subset of stars of those cmds are represented , to equal number of stars in the cmd of qj0033 , the distribution of stars around the different isochrones is very similar . in these three fields
, the bulk of stars seems to be older than 3 - 4 gyr .
a similar conclusion was reached by dolphin et al .
( 2001 ) for a field located in the northwestern part and relatively close to field qj0033 ( see figure [ smchi ] ) , where they found a greatly decreased star formation in the past 2 gyr . regarding the fields observed toward the south , the young population is less important from r=2.7@xmath36 on .
the blue elevation in the cfs of the southern fields is only found in the two fields closer to the center . in the case of smc0057 , its shape is consistent with some enhancement of the recent sfh as compared to a constant sfr(t ) . from the cfs
it is evident that field smc0100 has many more stars younger than 2 gyr ( relative to the total population of the field ) than the western fields qj0036 and qj0037 , even when it is almost 1@xmath36 further from the smc center than the latters . note that this field is located at the border of a more densely populated isopleth than qj0036 and qj0037 and in the part in which the trumpet - like distribution of the young population _
fans out_. beyond @xmath02.7@xmath36 , there are only few stars younger than 3 gyr old and the differences between the fields are small ( except for the fact that the density of stars in field smc0049 is larger than in the other two ) . finally , except for eastermost field qj0102 , the further south from the smc center we go , the more is the main peak of the cf shifted to the red .
this , together with the above facts , may indicate that the _ main _ epoch of star formation beyond @xmath02.2@xmath36 in this direction ended around 3 - 4 gyr ago .
the smc shows a clear morphologic dichotomy between its young and old population .
the youngest component has an asymmetric distribution which _ fans out _ towards the `` wing '' area in the northeast , similarly to the hi distribution ( stanimirovi et al .
1999 ) , while the rgb and agb stars have a more symmetric , spheroidal distribution . using deep ( down to @xmath46 ) cmds in 12 small fields that strategically sample different smc regions
, we are trying to shed some new light into our knowledge of the smc evolution , and answer some of the questions posed in the introduction .
we found that the underlying spheroidally distributed population is mainly composed by intermediate - age and old stars , and its distribution does not show strong galactocentric gradients .
our fields situated toward the northwest of the bar , at galactocentric distances from 1.3@xmath36 to 2.9@xmath36 contain very few stars younger than @xmath47 3 gyr .
the comparison of their cfs with that of a synthetic population with constant sfr(t ) indicates there has been an actual drop of the sfr(t ) at around this age until the present . even though , some residual star formation seems to have continued up to now .
no strong differences can be noticed between the cmds and cfs of the fields at 1.3@xmath36 and 1.4@xmath36 and the one at 2.9@xmath36 .
the apparent differences between these cmds are mainly due to the different number of stars in them , rather to a fundamentally different composition of the stellar populations .
something similar occurs in the case of the southern fields beyond @xmath48 .
in addition , we estimated that the sfr(t ) at intermediate to old ages in fields located at the same distance from the smc center and at either side ( e.g. qj0111 and qj0036 ) is similar .
with the current analysis of the data we can not provide a detailed sfh for these fields , and therefore we can not confirm nor rule out periods of quiescence as found in other studies that used shallower data .
this will be the subject of a forthcoming paper .
the three fields situated towards the east , in the `` wing '' region , show very active current star formation . in the case of the field closer to the center ,
qj0111 , we estimated that the current star formation rate ( from around 1 gyr ago to the present ) in this field is @xmath030% larger than the sfr(t ) averaged over the galaxy s life .
however , the other two northeastern fields , situated just slightly further away from the center , have a cf which is compatible with that of a constant sfr(t ) over the whole galaxy s history .
therefore , only the parts of the wing closer to the center , seem to be undergoing an exceptionally intense episode of star formation at the present time . at the radius of the innermost field ( @xmath01 kpc )
, we derived a crossing time of @xmath010@xmath49 yr , obtained adopting a measured velocity dispersion of 36 km / s ( carrera 2006 , using the stars below the rgb tip corresponding to our smc fields ) . in the case of our outermost field
( @xmath04 kpc ) the crossing time is @xmath02@xmath510@xmath49 yr .
large scale asymmetries should disappear within a few crossing times , and the population older than @xmath01 gyr is then expected to be well mixed at a radius of @xmath01 kpc , while the population older than @xmath02 gyr is well mixed at @xmath04 kpc . hence , although our fields only cover little isolated parts of the smc ( as compared with the huge area embraced by hz04 ) , they are representative of the smc populations older than @xmath02 gyr . the larger amount of young stars present in the eastern fields may be related to a recent interaction between the smc , the lmc , and the milky way as is suggested by numerical models ( e.g. yoshizawa & noguchi 2003 ) . note that the presence of a considerable young population in the eastern fields and lack thereof in the western ones is in good correspondence with the existence or absence of large amounts of hi at the corresponding locations ( stanimirovi et al .
it is interesting to notice that this trend is not maintained in the closest southern field , smc0057 , in which there is a considerable amount of young population but the amount of hi is small .
the presence of a substantial amount of intermediate - age population in our outermost smc fields , could be a sign that a few gigayears ago hi gas extended as far as @xmath04@xmath36 and has since gradually receded to the more central regions .
another possible explanation for this is that tidal interactions happened periodically , and then these stars dynamically mixed , resulting in old stars ( those from previous passages ) being symmetrically distributed and the younger ones only in the current tidal feature .
none of the studied fields is dominated by an old stellar component , as one would expect for an old stellar halo similar to the one of the milky way .
this may mean that a disk population is still dominating over a possible old halo , if it does exist in the smc .
we want to thank antonio aparicio and santi cassisi for many very useful opinions and discussions and a careful reading of the manuscript .
we are very grateful to the anonymous referee for a critical and useful report that helped to improve the paper .
the authors acknowledge support by the plan nacional de investigacin cientfica , desarrollo , e investigacin tecnolgica , ( aya2004 - 06343 ) .
e. c. and r. a. m. acknowledge support by the fondo nacional de investigacin cientfica y tecnolgica ( no .
1050718 , fondecyt ) and by the chilean centro de astrofsica fondap ( no . 15010003 ) .
this project has made generous use of the 10% chilean time , and continuous support from the cntac and las campanas staff is greatly appreciated .
aparicio , a. , & gallart , c. 2004 , , 128 , 1465 carrera , r.j .
2006 , phd thesis cioni , m .-
l. , habing , h. j. , israel , f. p. 2000 , a&a , 358 , 9l ct , s. , freeman , k. c. , carignan , c. & quinn , p. j. 1997 , , 114 , 1313 dolphin , a. e. , walker , a. r. , hodge , p. w. , mateo , m. , olszewski , e. w. , schommer , r. a. , & suntzeff , n. b. 2001 , , 562 , 303 ferguson , h. c. , & sandage , a. 1991 , , 101 , 765 gallart , c. 1998 , , 495 , 43l gallart , c. , freedman , w. l. , aparicio , a. , bertelli , g. 1999 , , 118 , 2245 gallart , c. 2000 , in asp conf .
221 , stars , gas and dust in galaxies : exploring the links , ed .
d. alloin , k. olsen , & g. galaz ( san francisco , asp ) , 133 gallart , c. , zoccali , m. , aparicio , a. 2005 , ara&a , 43 , 10 gardiner , l. t. , & hatzidimitriou , d. 1992 , , 257 , 195 girardi , l. , bertelli , g. , bressan , a. , chiosi , c. , groenewegen , m. a. t. , marigo , p. , salasnich , b. , & weiss , a. 2002 , a&a , 391 , 195 harris , j. , & zaritsky , d. 2004 , , 127 , 1531 ( hz04 ) holtzman , j. a. , gallagher , j. s. iii , cole , a. a. , et al .
1999 , , 118 , 2262 landolt , a. u. 1992 , , 104 , 340 madau , p. , pozzetti , l. , dickinson , m. 1998 , , 498 , 106 mccumber , m. p. , garnett , d. r. , & dufour , r. j. 2005 , , 130 , 1083 pedreros , m. h. , costa , e. , & mndez , r. a. 2006 , , 131 , 1461 pietrinferni , a. , cassisi , s. , salaris , m. , & castelli , f. 2004 , , 612 , 168 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 stanimirovi , s. , staveley - smith , l. , dickey , j. m. , sault , r. j. & snowden , s. l. 1999 , , 302 , 417 stetson , p. b. 1987 , , 99 , 191 stetson , p. b. 1990 , , 102 , 932 stetson , p. b. 1992 , in asp conf .
25 , astronomical data analysis software and systems i , ed .
d. m. worrall , c. biemesderfer , & j. barnes ( san francisco , asp ) , 297 stetson , p. b. 1993 , user s manual for daophot ii stetson , p. b. 1994 , , 106 , 250 stetson , p. b. , bruntt , h. , & grundahl , f. 2003 , , 115 , 413 tinney , c. g. , da costa , g. s. , & zinnecker , h. 1997 , , 285 , 111 van den bergh , s. 1999 , ara&a , 9 , 273 yoshizawa , a. m. , noguchi , m. 2003 , mnras , 339 , 1135 zaritsky , d. , harris , j. , grebel , e. k. , & thompson , i. b. 2000 , , 534 , 53l zaritsky , d. , harris , j. , & thompson , i. b. 1997 , , 114 , 1002 | we present ground - based _ b _ and _ r_-band color - magnitude diagrams ( cmds ) , reaching the oldest main - sequence turnoffs with good photometric accuracy for twelve fields in the small magellanic cloud ( smc ) . our fields , located between @xmath01@xmath1 and @xmath04@xmath1 from the center of the galaxy , are situated in different parts of the smc such as the `` wing '' area , and towards the west and south . in this paper we perform a first analysis of the stellar content in our smc fields through comparison with theoretical isochrones and color functions ( cfs ) .
we find that the underlying spheroidally distributed population is composed of both intermediate - age and old stars and that its age composition does not show strong galacto- centric gradients .
the three fields situated toward the east , in the wing region , show very active current star formation .
however , only in the eastern field closest to the center do we find an enhancement of recent star formation with respect to a constant sfr(t ) .
the fields corresponding to the western side of the smc present a much less populated young ms , and the cf analysis indicates that the sfr@xmath2 greatly diminished around 2 gyr ago in these parts .
field smc0057 , the closest to the center of the galaxy and located in the southern part , shows recent star formation , while the rest of the southern fields present few bright ms stars .
the structure of the red clump in all the cmds is consistent with the large amount of intermediate - age stars inferred from the cmds and color functions .
none of the smc fields presented here are dominated by old stellar populations , a fact that is in agreement with the lack of a conspicuous horizontal branch in all these smc cmds .
this could indicate that a disk population is ruling over a possible old halo in all the observed fields . |
information and technology make large data sets widely available for scientific discovery .
much statistical analysis of such high - dimensional data involves the estimation of a covariance matrix or its inverse ( the precision matrix ) .
examples include portfolio management and risk assessment ( fan , fan and lv , 2008 ) , high - dimensional classification such as fisher discriminant ( hastie , tibshirani and friedman , 2009 ) , graphic models ( meinshausen and bhlmann , 2006 ) , statistical inference such as controlling false discoveries in multiple testing ( leek and storey , 2008 ; efron , 2010 ) , finding quantitative trait loci based on longitudinal data ( yap , fan , and wu , 2009 ; xiong et al . 2011 ) , and testing the capital asset pricing model ( sentana , 2009 ) , among others .
see section 5 for some of those applications .
yet , the dimensionality is often either comparable to the sample size or even larger . in such cases ,
the sample covariance is known to have poor performance ( johnstone , 2001 ) , and some regularization is needed .
realizing the importance of estimating large covariance matrices and the challenges brought by the high dimensionality , in recent years researchers have proposed various regularization techniques to consistently estimate @xmath0 .
one of the key assumptions is that the covariance matrix is sparse , namely , many entries are zero or nearly so ( bickel and levina , 2008 , rothman et al , 2009 , lam and fan 2009 , cai and zhou , 2010 , cai and liu , 2011 ) . in many applications
, however , the sparsity assumption directly on @xmath0 is not appropriate .
for example , financial returns depend on the equity market risks , housing prices depend on the economic health , gene expressions can be stimulated by cytokines , among others . due to the presence of common factors , it is unrealistic to assume that many outcomes are uncorrelated .
an alternative method is to assume a factor model structure , as in fan , fan and lv ( 2008 ) .
however , they restrict themselves to the strict factor models with known factors .
a natural extension is the conditional sparsity .
given the common factors , the outcomes are weakly correlated .
in order to do so , we consider an approximate factor model , which has been frequently used in economic and financial studies ( chamberlain and rothschild , 1983 ; fama and french 1993 ; bai and ng , 2002 , etc ) : @xmath1 here @xmath2 is the observed response for the @xmath3th ( @xmath4 ) individual at time @xmath5 ; @xmath6 is a vector of factor loadings ; @xmath7 is a @xmath8 vector of common factors , and @xmath9 is the error term , usually called _
idiosyncratic component _ , uncorrelated with @xmath7 . both @xmath10 and @xmath11
diverge to infinity , while @xmath12 is assumed fixed throughout the paper , and @xmath10 is possibly much larger than @xmath11 .
we emphasize that in model ( [ eq1.1 ] ) , only @xmath2 is observable .
it is intuitively clear that the unknown common factors can only be inferred reliably when there are sufficiently many cases , that is , @xmath13 . in a data - rich environment
, @xmath10 can diverge at a rate faster than @xmath11 .
the factor model ( [ eq1.1 ] ) can be put in a matrix form as @xmath14 where @xmath15 , @xmath16 and @xmath17 .
we are interested in @xmath0 , the @xmath18 covariance matrix of @xmath19 , and its inverse , which are assumed to be time - invariant . under model ( [ eq1.1 ] ) , @xmath0 is given by @xmath20 where @xmath21 is the covariance matrix of @xmath22 .
the literature on approximate factor models typically assumes that the first @xmath12 eigenvalues of @xmath23 diverge at rate @xmath24 , whereas all the eigenvalues of @xmath25 are bounded as @xmath26 .
this assumption holds easily when the factors are pervasive in the sense that a non - negligible fraction of factor loadings should be non - vanishing .
the decomposition ( [ eq1.3 ] ) is then asymptotically identified as @xmath26 .
in addition to it , in this paper we assume that @xmath25 is _ approximately sparse _ as in bickel and levina ( 2008 ) and rothman et al .
( 2009 ) : for some @xmath27 , @xmath28 does not grow too fast as @xmath29 in particular , this includes the exact sparsity assumption ( @xmath30 ) under which @xmath31 , the maximum number of nonzero elements in each row . the conditional sparsity structure of ( [ eq1.2 ] )
was explored by fan , liao and mincheva ( 2011 ) in estimating the covariance matrix , when the factors @xmath32 are observable .
this allows them to use regression analysis to estimate @xmath33 .
this paper deals with the situation in which the factors are unobservable and have to be inferred .
our approach is simple , optimization - free and it uses the data only through the sample covariance matrix .
run the singular value decomposition on the sample covariance matrix @xmath34 of @xmath19 , keep the covariance matrix formed by the first @xmath12 principal components , and apply the thresholding procedure to the remaining covariance matrix .
this results in a principal orthogonal complement thresholding ( poet ) estimator .
when the number of common factors @xmath12 is unknown , it can be estimated from the data .
see section 2 for additional details .
we will investigate various properties of poet under the assumption that the data are serially dependent , which includes independent observations as a specific example .
the rate of convergence under various norms for both estimated @xmath0 and @xmath25 and their precision ( inverse ) matrices will be derived .
we show that the effect of estimating the unknown factors on the rate of convergence vanishes when @xmath35 , and in particular , the rate of convergence for @xmath25 achieves the optimal rate in cai and zhou ( 2012 ) .
this paper focuses on the high - dimensional _ static factor model _ ( [ eq1.2 ] ) , which is innately related to the principal component analysis ( pca ) , as clarified in section 2 .
this feature makes it different from the classical factor model with fixed dimensionality ( e.g. , lawley and maxwell 1971 ) . in the last ten years
, much theory on the estimation and inference of the static factor model has been developed , for example , stock and watson ( 1998 , 2002 ) , bai and ng ( 2002 ) , bai ( 2003 ) , doz , giannone and reichlin ( 2011 ) , among others .
our contribution is on the estimation of covariance matrices and their inverse in large factor models .
the _ static _ model considered in this paper is to be distinguished from the _ dynamic factor model _ as in forni , hallin , lippi and reichlin ( 2000 ) ; the latter allows @xmath19 to also depend on @xmath7 with lags in time .
their approach is based on the eigenvalues and principal components of spectral density matrices , and on the frequency domain analysis .
moreover , as shown in forni and lippi ( 2001 ) , the dynamic factor model does not really impose a restriction on the data generating process , and the assumption of idiosyncrasy ( in their terminology , a @xmath10-dimensional process is idiosyncratic if all the eigenvalues of its spectral density matrix remain bounded as @xmath26 ) asymptotically identifies the decomposition of @xmath2 into the common component and idiosyncratic error .
the literature includes , for example , forni et al .
( 2000 , 2004 ) , forni and lippi ( 2001 ) , hallin and lika ( 2007 , 2011 ) , and many other references therein .
above all , both the static and dynamic factor models are receiving increasing attention in applications of many fields where information usually is scattered through a ( very ) large number of interrelated time series .
there has been extensive literature in recent years that deals with sparse principal components , which has been widely used to enhance the convergence of the principal components in high - dimensional space .
daspremont , bach and el ghaoui ( 2008 ) , shen and huang ( 2008 ) , witten , tibshirani , and hastie ( 2009 ) and ma ( 2011 ) proposed and studied various algorithms for computations . more literature on sparse pca is found in johnstone and lu ( 2009 ) , amini and wainwright ( 2009 ) , zhang and el ghaoui ( 2011 ) , birnbaum et al .
( 2012 ) , among others .
in addition , there has also been a growing literature that theoretically studies the recovery from a low - rank plus sparse matrix estimation problem , see for example , wright et al .
( 2009 ) , lin et al .
( 2009 ) , cands et al .
( 2011 ) , luo ( 2011 ) , agarwal , nagahban , wainwright ( 2012 ) , pati et al .
it corresponds to the identifiability issue of our problem .
there is a big difference between our model and those considered in the aforementioned literature . in the current paper ,
the first @xmath12 eigenvalues of @xmath0 are spiked and grow at a rate @xmath24 , whereas the eigenvalues of the matrices studied in the existing literature on covariance estimation are usually assumed to be either bounded or slowly growing . due to this distinctive feature , the common components and the idiosyncratic components can be identified , and in addition , pca on the sample covariance matrix can consistently estimate the space spanned by the eigenvectors of @xmath0 .
the existing methods of either thresholding directly or solving a constrained optimization method can fail in the presence of very spiked principal eigenvalues .
however , there is a price to pay here : as the first @xmath36 eigenvalues are too spiked " , one can hardly obtain a satisfactory rate of convergence for estimating @xmath0 in absolute term , but it can be estimated accurately in relative term ( see section 3.3 for details ) .
in addition , @xmath37 can be estimated accurately .
we would like to further note that the low - rank plus sparse representation of our model is on the population covariance matrix , whereas cands et al .
( 2011 ) , wright et al .
( 2009 ) , lin et al .
( 2009 ) considered such a representation on the data matrix .
as there is no @xmath0 to estimate , their goal is limited to producing a low - rank plus sparse matrix decomposition of the data matrix , which corresponds to the identifiability issue of our study , and does not involve estimation and inference .
in contrast , our ultimate goal is to estimate the population covariance matrices as well as the precision matrices . for this purpose
, we require the idiosyncratic components and common factors to be uncorrelated and the data generating process to be strictly stationary .
the covariances considered in this paper are constant over time , though slow - time - varying covariance matrices are applicable through localization in time ( time - domain smoothing ) .
our consistency result on @xmath25 demonstrates that the decomposition ( [ eq1.3 ] ) is identifiable , and hence our results also shed the light of the surprising phenomenon " of cands et al .
( 2011 ) that one can separate fully a sparse matrix from a low - rank matrix when only the sum of these two components is available .
the rest of the paper is organized as follows .
section 2 gives our estimation procedures and builds the relationship between the principal components analysis and the factor analysis in high - dimensional space .
section 3 provides the asymptotic theory for various estimated quantities .
section 4 illustrates how to choose the thresholds using cross - validation and guarantees the positive definiteness in any finite sample .
specific applications of regularized covariance matrices are given in section 5 .
numerical results are reported in section 6 .
finally , section 7 presents a real data application on portfolio allocation .
all proofs are given in the appendix . throughout the paper ,
we use @xmath38 and @xmath39 to denote the minimum and maximum eigenvalues of a matrix @xmath40 .
we also denote by @xmath41 , @xmath42 , @xmath43 and @xmath44 the frobenius norm , spectral norm ( also called operator norm ) , @xmath45-norm , and elementwise norm of a matrix @xmath40 , defined respectively by @xmath46 , @xmath47 , @xmath48 and @xmath49 .
note that when @xmath40 is a vector , both @xmath41 and @xmath42 are equal to the euclidean norm .
finally , for two sequences , we write @xmath50 if @xmath51 and @xmath52 if @xmath53 and @xmath54
there are three main objectives of this paper : ( i ) understand the relationship between principal component analysis ( pca ) and the high - dimensional factor analysis ; ( ii ) estimate both covariance matrices @xmath0 and the idiosyncratic @xmath25 and their precision matrices in the presence of common factors , and ( iii ) investigate the impact of estimating the unknown factors on the covariance estimation . the propositions in section [ s2.1 ] below show that the space spanned by the principal components in the population level @xmath0 is close to the space spanned by the columns of the factor loading matrix @xmath55 .
consider a factor model @xmath56 where the number of common factors , @xmath57 , is small compared to @xmath10 and @xmath11 , and thus is assumed to be fixed throughout the paper . in the model , the only observable variable is the data @xmath2 .
one of the distinguished features of the factor model is that the principal eigenvalues of @xmath0 are no longer bounded , but growing fast with the dimensionality .
we illustrate this in the following example .
consider a single - factor model @xmath58 where @xmath59 suppose that the factor is pervasive in the sense that it has non - negligible impact on a non - vanishing proportion of outcomes .
it is then reasonable to assume @xmath60 for some @xmath61 .
therefore , assuming that @xmath62 , an application of ( [ eq1.3 ] ) yields , @xmath63 for all large @xmath10 , assuming @xmath64 .
we now elucidate why pca can be used for the factor analysis in the presence of spiked eigenvalues .
write @xmath65 as the @xmath66 loading matrix .
note that the linear space spanned by the first @xmath12 principal components of @xmath23 is the same as that spanned by the columns of @xmath55 when @xmath67 is non - degenerate .
thus , we can assume without loss of generality that the columns of @xmath55 are orthogonal and @xmath68 , the identity matrix .
this canonical form corresponds to the identifiability condition in decomposition ( [ eq1.3 ] ) .
let @xmath69 be the columns of @xmath55 , ordered such that @xmath70 is in a non - increasing order .
then , @xmath71 are eigenvectors of the matrix @xmath72 with eigenvalues @xmath73 and the rest zero .
we will impose the pervasiveness assumption that all eigenvalues of the @xmath74 matrix @xmath75 are bounded away from zero , which holds if the factor loadings @xmath76 are independent realizations from a non - degenerate population .
since the non - vanishing eigenvalues of the matrix @xmath72 are the same as those of @xmath77 , from the pervasiveness assumption it follows that @xmath78 are all growing at rate @xmath24 .
let @xmath79 be the eigenvalues of @xmath80 in a descending order and @xmath81 be their corresponding eigenvectors .
then , an application of weyl s eigenvalue theorem ( see the appendix ) yields that [ prop21 ] assume that the eigenvalues of @xmath75 are bounded away from zero for all large @xmath10 . for the factor model ( [ eq1.3 ] ) with the canonical condition @xmath82 we have @xmath83 in addition , for @xmath84 , @xmath85 . using proposition [ prop21 ] and the @xmath86 theorem of davis and kahn ( 1970 , see the appendix )
, we have the following : [ prop22 ] under the assumptions of proposition [ prop21 ] , if @xmath70 are distinct , then @xmath87 propositions [ prop21 ] and [ prop22 ] state that pca and factor analysis are approximately the same if @xmath88 .
this is assured through a sparsity condition on @xmath21 , which is frequently measured through @xmath89$.}\ ] ] the intuition is that , after taking out the common factors , many pairs of the cross - sectional units become weakly correlated .
this generalized notion of sparsity was used in bickel and levina ( 2008 ) and cai and liu ( 2011 ) . under this generalized measure of sparsity
, we have @xmath90 if the noise variances @xmath91 are bounded .
therefore , when @xmath92 , proposition [ prop21 ] implies that we have distinguished eigenvalues between the principal components @xmath93 and the rest of the components @xmath94 and proposition [ prop22 ] ensures that the first @xmath12 principal components are approximately the same as the columns of the factor loadings .
the aforementioned sparsity assumption appears reasonable in empirical applications .
boivin and ng ( 2006 ) conducted an empirical study and showed that imposing zero correlation between weakly correlated idiosyncratic components improves forecast .
more recently , phan ( 2012 ) empirically estimated the level of sparsity of the idiosyncratic covariance using the uk market data .
recent developments on random matrix theory , for example , johnstone and lu ( 2009 ) and paul ( 2007 ) , have shown that when @xmath95 is not negligible , the eigenvalues and eigenvectors of @xmath0 might not be consistently estimated from the sample covariance matrix .
a distinguished feature of the covariance considered in this paper is that there are some very spiked eigenvalues . by propositions 2.1 and 2.2 , in the factor model
, the pervasiveness condition @xmath96 implies that the first @xmath12 eigenvalues are growing at a rate @xmath10 .
moreover , when @xmath10 is large , the principal components @xmath97 are close to the normalized vectors @xmath98 when @xmath99 .
this provides the mathematics for using the first @xmath12 principal components as a proxy of the space spanned by the columns of the factor loading matrix @xmath55 .
in addition , due to ( [ e2.4 ] ) , the signals of the first @xmath12 eigenvalues are stronger than those of the spiked covariance model considered by jung and marron ( 2009 ) and birnbaum et al .
therefore , our other conditions for the consistency of principal components at the population level are much weaker than those in the spiked covariance literature . on the other hand , this also shows that , under our setting the pca is a valid approximation to factor analysis only if @xmath26 . the fact that the pca on the sample covariance is inconsistent when
@xmath10 is bounded was also previously demonstrated in the literature ( see e.g. , bai ( 2003 ) ) . with assumption ( [ e2.4 ] )
, the standard literature on approximate factor models has shown that the pca on the sample covariance matrix @xmath34 can consistently estimate the space spanned by the factor loadings ( e.g. , stock and watson ( 1998 ) , bai ( 2003 ) ) .
our contribution in propositions 2.1 and 2.2 is that we connect the high - dimensional factor model to the principal components , and obtain the consistency of the spectrum in the population level @xmath0 instead of the sample level @xmath34 .
the spectral consistency also enhances the results in chamberlain and rothschild ( 1983 ) .
this provides the rationale behind the consistency results in the factor model literature .
sparsity assumption directly on @xmath0 is inappropriate in many applications due to the presence of common factors .
instead , we propose a nonparametric estimator of @xmath0 based on the principal component analysis .
let @xmath100 be the ordered eigenvalues of the sample covariance matrix @xmath101 and @xmath102 be their corresponding eigenvectors .
then the sample covariance has the following spectral decomposition : @xmath103 where @xmath104 is the principal orthogonal complement , and @xmath12 is the number of diverging eigenvalues of @xmath0 . let us first assume @xmath12 is known .
now we apply thresholding on @xmath105 .
define @xmath106 where @xmath107 is a generalized shrinkage function of antoniadis and fan ( 2001 ) , employed by rothman et al .
( 2009 ) and cai and liu ( 2011 ) , and @xmath108 is an entry - dependent threshold . in particular , the hard - thresholding rule @xmath109 ( bickel and levina , 2008 ) and the constant thresholding parameter @xmath110 are allowed . in practice
, it is more desirable to have @xmath111 be entry - adaptive .
an example of the adaptive thresholding is @xmath112 where @xmath113 is the @xmath114 diagonal element of @xmath105 .
this corresponds to applying the thresholding with parameter @xmath115 to the correlation matrix of @xmath105 .
the estimator of @xmath0 is then defined as : @xmath116 we will call this estimator the principal orthogonal complement thresholding ( poet ) estimator .
it is obtained by thresholding the remaining components of the sample covariance matrix , after taking out the first @xmath12 principal components .
one of the attractiveness of poet is that it is optimization - free , and hence is computationally appealing . ,
@xmath25 , @xmath12 , the factors and loadings . ] with the choice of @xmath111 in ( [ eq2.3 ] ) and the hard thresholding rule , our estimator encompasses many popular estimators as its specific cases .
when @xmath117 , the estimator is the sample covariance matrix and when @xmath118 , the estimator becomes that based on the strict factor model ( fan , fan , and lv , 2008 ) .
when @xmath119 , our estimator is the same as the thresholding estimator of bickel and levina ( 2008 ) and ( with a more general thresholding function ) rothman et al .
( 2009 ) or the adaptive thresholding estimator of cai and liu ( 2011 ) with a proper choice of @xmath111 . in practice ,
the number of diverging eigenvalues ( or common factors ) can be estimated based on the sample covariance matrix .
determining @xmath12 in a data - driven way is an important topic , and is well understood in the literature .
we will describe the poet with a data - driven @xmath12 in section [ s2.4 ] .
the poet ( [ eq2.4 ] ) has an equivalent representation using a constrained least squares method .
the least squares method seeks for @xmath120 and @xmath121 such that @xmath122 subject to the normalization @xmath123 the constraints ( [ eq2.11 ] ) correspond to the normalization ( [ eq2.7 ] ) . here
we assume that the mean of each variable @xmath124 has been removed , that is , @xmath125 for all @xmath126 and @xmath127 putting it in a matrix form , the optimization problem can be written as @xmath128 where @xmath129 and @xmath130 .
for each given @xmath131 , the least - squares estimator of @xmath55 is @xmath132 , using the constraint ( [ eq2.11 ] ) on the factors .
substituting this into ( [ eq2.12 ] ) , the objective function now becomes @xmath133 .
$ ] the minimizer is now clear : the columns of @xmath134 are the eigenvectors corresponding to the @xmath12 largest eigenvalues of the @xmath135 matrix @xmath136 and @xmath137 ( see e.g. , stock and watson ( 2002 ) )
. we will show that under some mild regularity conditions , as @xmath10 and @xmath138 , @xmath139 consistently estimates the true @xmath140 uniformly over @xmath141 and @xmath142 . since @xmath25 is assumed to be sparse , we can construct an estimator of @xmath25 using the adaptive thresholding method by cai and liu ( 2011 ) as follows .
let @xmath143 and @xmath144 for some pre - determined decreasing sequence @xmath145 , and large enough @xmath146 , define the adaptive threshold parameter as @xmath147 the estimated idiosyncratic covariance estimator is then given by @xmath148 where for all @xmath149 ( see antoniadis and fan , 2001 ) , @xmath150 it is easy to verify that @xmath107 includes many interesting thresholding functions such as the hard thresholding ( @xmath151 ) , soft thresholding ( @xmath152 ) , scad , and adaptive lasso ( see rothman et al .
( 2009 ) ) .
analogous to the decomposition ( [ eq1.3 ] ) , we obtain the following substitution estimators @xmath153 and by the sherman - morrison - woodbury formula , noting that @xmath154 @xmath155^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}_k'({\widehat{\mathbf{\sigma}}}_{u , k}^{\mathcal{t}})^{-1},\ ] ] in practice , the true number of factors @xmath12 might be unknown to us .
however , for any determined @xmath156 , we can always construct either @xmath157 as in ( [ eq2.4 ] ) or @xmath158 as in ( [ eq2.14 ] ) to estimate @xmath159 .
the following theorem shows that for each given @xmath160 , the two estimators based on either regularized pca or least squares substitution are equivalent .
similar results were obtained by bai ( 2003 ) when @xmath161 and no thresholding was imposed .
[ thm2.1 ] suppose that the entry - dependent threshold in ( [ eq2.2 ] ) is the same as the thresholding parameter used in ( [ eq2.13 ] ) .
then for any @xmath156 , the estimator ( [ eq2.4 ] ) is equivalent to the substitution estimator ( [ eq2.14 ] ) , that is , @xmath162 in this paper , we will use a data - driven @xmath163 to construct the poet ( see section 2.4 below ) , which has two equivalent representations according to theorem [ thm2.1 ] . determining the number of factors in a data - driven way has been an important research topic in the econometric literature .
bai and ng ( 2002 ) proposed a consistent estimator as both @xmath10 and @xmath11 diverge .
other recent criteria are proposed by kapetanios ( 2010 ) , onatski ( 2010 ) , alessi et al .
( 2010 ) , etc .
our method also allows a data - driven @xmath163 to estimate the covariance matrices . in principle , any procedure that gives a consistent estimate of @xmath12 can be adopted . in this paper
we apply the well - known method in bai and ng ( 2002 ) .
it estimates @xmath12 by @xmath164 where @xmath165 is a prescribed upper bound , @xmath166 is a @xmath167 matrix whose columns are @xmath168 times the eigenvectors corresponding to the @xmath160 largest eigenvalues of the @xmath135 matrix @xmath169 ; @xmath170 is a penalty function of @xmath171 such that @xmath172 and @xmath173 two examples suggested by bai and ng ( 2002 ) are @xmath174 @xmath175 throughout the paper , we let @xmath163 be the solution to ( [ eq2.16add ] ) using either ic1 or ic2 .
the asymptotic results are not affected regardless of the specific choice of @xmath170 .
we define the poet estimator with unknown @xmath12 as @xmath176 the procedure is as stated in section [ s2.2 ] except that @xmath177 is now data - driven .
this section presents the assumptions on the model ( [ eq1.2 ] ) , in which only @xmath178 are observable .
recall the identifiability condition ( [ eq2.7 ] ) .
the first assumption has been one of the most essential ones in the literature of approximate factor models . under this assumption and other regularity conditions ,
the number of factors , loadings and common factors can be consistently estimated ( e.g. , stock and watson ( 1998 , 2002 ) , bai and ng ( 2002 ) , bai ( 2003 ) , etc . ) .
[ a35 ] all the eigenvalues of the @xmath74 matrix @xmath179 are bounded away from both zero and infinity as @xmath26 . 1 .
it implies from proposition 2.1 in section 2 that the first @xmath12 eigenvalues of @xmath0 grow at rate @xmath24 .
this unique feature distinguishes our work from most of other low - rank plus sparse covariances considered in the literature , e.g. , luo ( 2011 ) , pati et al .
( 2012 ) , agarwal et al .
( 2012 ) , birnbaum et al .
( 2012 ) . )
are fan et al .
( 2008 , 2011 ) and bai and shi ( 2011 ) . while fan et al .
( 2008 , 2011 ) assumed the factors are observable , bai and shi ( 2011 ) considered the strict factor model in which @xmath25 is diagonal . ]
assumption 3.1 requires the factors to be pervasive , that is , to impact a non - vanishing proportion of individual time series .
see example 2.1 for its meaning .
is sparse the intuition of a sparse loading matrix is that each factor is related to only a relatively small number of stocks , assets , genes , etc . with @xmath55 being sparse
, all the eigenvalues of @xmath77 and hence those of @xmath0 are bounded . ]
3 . as to be illustrated in section 3.3 below , due to the fast diverging eigenvalues
, one can hardly achieve a good rate of convergence for estimating @xmath0 under either the spectral norm or frobenius norm when @xmath180 .
this phenomenon arises naturally from the characteristics of the high - dimensional factor model , which is another distinguished feature compared to those convergence results in the existing literature .
[ a21 ] ( i ) @xmath181 is strictly stationary .
in addition , @xmath182 for all @xmath126 and @xmath127 + ( ii ) there exist constants @xmath183 such that @xmath184 , @xmath185 and @xmath186 + ( iii ) there exist @xmath187 and @xmath188 , such that for any @xmath189 , @xmath141 and @xmath84 , @xmath190 condition ( i ) requires strict stationarity as well as the non - correlation between @xmath191 and @xmath32 .
these conditions are slightly stronger than those in the literature , e.g. , bai ( 2003 ) , but are still standard and simplify our technicalities .
condition ( ii ) requires that @xmath25 be well - conditioned .
the condition @xmath192 instead of a weaker condition @xmath193 is imposed here in order to consistently estimate @xmath12 .
but it is still standard in the approximate factor model literature as in bai and ng ( 2002 ) , bai ( 2003 ) , etc .
when @xmath12 is known , such a condition can be removed .
our working paper shows that the results continue to hold for a growing ( known ) @xmath12 under the weaker condition @xmath193 . condition ( iii ) requires exponential - type tails , which allows us to apply the large deviation theory to @xmath194 and @xmath195 .
we impose the strong mixing condition .
let @xmath196 and @xmath197 denote the @xmath198-algebras generated by @xmath199 and @xmath200 respectively .
in addition , define the mixing coefficient @xmath201 [ a32 ] strong mixing : there exists @xmath202 such that @xmath203 , and @xmath146 satisfying : for all @xmath204 , @xmath205 in addition , we impose the following regularity conditions .
[ a33 ] there exists @xmath206 such that for all @xmath141 , @xmath142 and @xmath207 , + ( i ) @xmath208 , + ( ii ) @xmath209 ^ 4<m$ ] , + ( iii ) @xmath210 .
these conditions are needed to consistently estimate the transformed common factors as well as the factor loadings .
similar conditions were also assumed in bai ( 2003 ) , and bai and ng ( 2006 ) .
the number of factors is assumed to be fixed .
our conditions in assumption [ a33 ] are weaker than those in bai ( 2003 ) as we focus on different aspects of the study . estimating the covariance matrix @xmath25 of the idiosyncratic components @xmath211 is important for many statistical inferences .
for example , it is needed for large sample inference of the unknown factors and their loadings , for testing the capital asset pricing model ( sentana , 2009 ) , and large - scale hypothesis testing ( fan , han and gu , 2012 ) . see section 5 .
we estimate @xmath25 by thresholding the principal orthogonal complements after the first @xmath163 principal components of the sample covariance are taken out : @xmath212 by theorem [ thm2.1 ] , it also has an equivalent expression given by ( [ eq2.13 ] ) , with @xmath213 . throughout the paper ,
we apply the adaptive threshold @xmath214 where @xmath146 is a sufficiently large constant , though the results hold for other types of thresholding . as in bickel and levina ( 2008 ) and
cai and liu ( 2011 ) , the threshold chosen in the current paper is in fact obtained from the optimal uniform rate of convergence of @xmath215 when direct observation of @xmath9 is not available , the effect of estimating the unknown factors also contributes to this uniform estimation error , which is why @xmath216 appears in the threshold .
the following theorem gives the rate of convergence of the estimated idiosyncratic covariance .
let @xmath217 . in the convergence rate below , recall that @xmath218 and @xmath219 are defined in the measure of sparsity ( [ eq2.5 ] ) .
[ thm31 ] suppose @xmath220 , @xmath221 , and assumptions [ a35]-[a33 ] hold . then for a sufficiently large constant @xmath146 in the threshold ( [ eq3.3 ] ) , the poet estimator @xmath222 satisfies @xmath223 if further @xmath224 , then the eigenvalues of @xmath222 are all bounded away from zero with probability approaching one , and @xmath225 when estimating @xmath25 , @xmath10 is allowed to grow exponentially fast in @xmath11 , and @xmath222 can be made consistent under the spectral norm .
in addition , @xmath222 is asymptotically invertible while the classical sample covariance matrix based on the residuals is not when @xmath226 [ rem3.3 ] 1 .
consistent estimation of @xmath25 indicates that @xmath25 is identifiable in ( [ eq1.3 ] ) , namely , the sparse @xmath25 can be separated perfectly from the low - rank matrix there .
the result here gives another proof ( when assuming @xmath227 ) of the surprising phenomenon " in cands et al ( 2011 ) under different technical conditions .
2 . fan , liao and mincheva ( 2011 ) recently showed that when @xmath228 are observable and @xmath30 , the rate of convergence of the adaptive thresholding estimator is given by @xmath229 hence when the common factors are unobservable , the rate of convergence has an additional term @xmath230 , coming from the impact of estimating the unknown factors .
this impact vanishes when @xmath231 , in which case the minimax rate as in cai and zhou ( 2010 ) is achieved .
as @xmath10 increases , more information about the common factors is collected , which results in more accurate estimation of the common factors @xmath228 .
3 . when @xmath12 is known and grows with @xmath10 and @xmath11 , with slightly weaker assumptions , our working paper ( fan et al .
2011 ) shows that under the exactly sparse case ( that is , @xmath30 ) , the result continues to hold with convergence rate @xmath232 . since the first @xmath12 eigenvalues of @xmath0 grow with @xmath10 , one can hardly estimate @xmath0 with satisfactory accuracy in the absolute term .
this problem arises not from the limitation of any estimation method , but is due to the nature of the high - dimensional factor model .
we illustrate this using a simple example .
[ exam31 ] consider an ideal case where we know the spectrum except for the first eigenvector of @xmath0 .
let @xmath233 be the eigenvalues and vectors , and assume that the largest eigenvalue @xmath234 for some @xmath61 .
let @xmath235 be the estimated first eigenvector and define the covariance estimator @xmath236 assume that @xmath235 is a good estimator in the sense that @xmath237 .
however , @xmath238 which can diverge when @xmath239 .
@xmath240 in the presence of very spiked eigenvalues , while the covariance @xmath0 can not be consistently estimated in absolute term , it can be well estimated in terms of the _ relative error _
matrix @xmath241 which is more relevant for many applications ( see example 5.2 ) .
the relative error matrix can be measured by either its spectral norm or the normalized frobenius norm defined by @xmath242\right)^{1/2}.\ ] ] in the last equality , there are @xmath10 terms being added in the trace operation and the factor @xmath243 plays the role of normalization . the loss ( [ eq.fan ] ) is closely related to the entropy loss , introduced by james and stein ( 1961 ) .
also note that @xmath244 where @xmath245 is the weighted quadratic norm in fan et al ( 2008 ) .
fan et al .
( 2008 ) showed that in a large factor model , the sample covariance is such that @xmath246 which does not converge if @xmath180 . on the other hand ,
theorem [ thm32 ] below shows that @xmath247 can still be convergent as long as @xmath248 . technically , the impact of high - dimensionality on the convergence rate of @xmath249 is via the number of rows in @xmath55 .
we show in the appendix that @xmath55 appears in @xmath247 through @xmath250 whose eigenvalues are bounded .
therefore it successfully cancels out the curse of high - dimensionality introduced by @xmath55 .
compared to estimating @xmath0 , in a large approximate factor model , we can estimate the precision matrix with a satisfactory rate under the spectral norm .
the intuition follows from the fact that @xmath37 has bounded eigenvalues .
the following theorem summarizes the rate of convergence under various norms .
[ thm32 ] under the assumptions of theorem [ thm31 ] , the poet estimator defined in ( [ eq2.16 ] ) satisfies @xmath251 in addition , if @xmath252 , then @xmath253 is nonsingular with probability approaching one , with @xmath254 1 . when estimating @xmath37 , @xmath10 is allowed to grow exponentially fast in @xmath11 , and the estimator has the same rate of convergence as that of the estimator @xmath222 in theorem [ thm31 ] .
when @xmath10 becomes much larger than @xmath11 , the precision matrix can be estimated at the same rate as if the factors were observable .
2 . as in remark
[ rem3.3 ] , when @xmath255 is known and grows with @xmath10 and @xmath11 , the working paper fan et al .
( 2011 ) proves the following results ( when @xmath30 ) : instead of @xmath256 be bounded . ]
@xmath257 the results state explicitly the dependence of the rate of convergence on the number of factors . 3 .
the relative error @xmath258 in operator norm can be shown to have the same order as the maximum relative error of estimated eigenvalues .
it does not converge to zero nor diverge .
it is much smaller than @xmath259 , which is of order @xmath260 ( see example [ exam31 ] ) .
many applications of the factor model require estimating the unknown factors . in general , factor loadings in @xmath261 and the common factors
@xmath262 are not separably identifiable , as for any matrix @xmath263 such that @xmath264 , @xmath265 . hence @xmath266
can not be identified from @xmath267 .
note that the linear space spanned by the rows of @xmath55 is the same as that by those of @xmath268 .
in practice , it often does not matter which one is used .
let @xmath269 denote the @xmath270 diagonal matrix of the first @xmath163 largest eigenvalues of the sample covariance matrix in decreasing order . recall that @xmath271 and define a @xmath270 matrix @xmath272 then for @xmath142 , @xmath273 note that @xmath274 depends only on the data @xmath275 and an identifiable part of parameters @xmath276 .
therefore , there is no identifiability issue in @xmath274 regardless of the imposed identifiability condition .
bai ( 2003 ) obtained the rate of convergence for both @xmath277 and @xmath278 for any fixed @xmath279 .
however , the uniform rate of convergence is more relevant for many applications ( see example 5.1 ) .
the following theorem extends those results in bai ( 2003 ) in a uniformity sense .
in particular , with a more refined technique , we have improved the uniform convergence rate for @xmath278 .
[ thm33 ] under the assumptions of theorem [ thm31 ] , @xmath280 as a consequence of theorem [ thm33 ] , we obtain the following : ( recall that the constant @xmath281 is defined in assumption [ a21 ] . ) [ c31 ] under the assumptions of theorem [ thm31 ] , @xmath282
the rates of convergence obtained above also explain the condition @xmath283 in theorems [ thm31 ] and [ thm32 ] .
it is needed in order to estimate the common factors @xmath228 uniformly in @xmath142 .
when we do not observe @xmath228 , in addition to the factor loadings , there are @xmath284 factors to estimate .
intuitively , the condition @xmath283 requires the number of parameters introduced by the unknown factors be not too many " , so that we can consistently estimate them uniformly .
technically , as demonstrated by bickel and levina ( 2008 ) , cai and liu ( 2011 ) and many other authors , achieving uniform accuracy is essential for large covariance estimations .
recall that the threshold value @xmath285 , where @xmath286 is determined by the users . to make poet operational in practice
, one has to choose @xmath286 to maintain the positive definiteness of the estimated covariances for any given finite sample .
we write @xmath287 , where the covariance estimator depends on @xmath286 via the threshold .
we choose @xmath286 in the range where @xmath288 .
define @xmath289 when @xmath286 is sufficiently large , the estimator becomes diagonal , while its minimum eigenvalue must retain strictly positive .
thus , @xmath290 is well defined and for all @xmath291 , @xmath292 is positive definite under finite sample .
we can obtain @xmath290 by solving @xmath293 we can also approximate @xmath290 by plotting @xmath294 as a function of @xmath286 , as illustrated in figure [ mineig ] . in practice , we can choose @xmath286 in the range @xmath295 for a small @xmath296 and large enough @xmath297 choosing the threshold in a range to guarantee the finite - sample positive definiteness has also been previously suggested by fryzlewicz ( 2012 ) . as a function of @xmath286 for three choices of thresholding rules .
the plot is based on the simulated data set in section 6.2.,title="fig:",width=302 ] [ mineig ] in practice , @xmath286 can be data - driven , and chosen through multifold cross - validation . after obtaining the estimated residuals @xmath298 by the pca , we divide them randomly into two subsets , which are , for simplicity , denoted by @xmath299 and @xmath300 .
the sizes of @xmath301 and @xmath302 , denoted by @xmath303 and @xmath304 , are @xmath305 and @xmath306 for example , in sparse matrix estimation , bickel and levina ( 2008 ) suggested to choose @xmath307 .
we repeat this procedure @xmath308 times . at the @xmath309th split ,
we denote by @xmath310 the poet estimator with the threshold @xmath311 on the training data set @xmath312 we also denote by @xmath313 the sample covariance based on the validation set , defined by @xmath314 then we choose the constant @xmath315 by minimizing a cross - validation objective function over a compact interval @xmath316 here @xmath290 is the minimum constant that guarantees the positive definiteness of @xmath292 for @xmath291 as described in the previous subsection , and @xmath165 is a large constant such that @xmath317 is diagonal .
the resulting @xmath315 is data - driven , so depends on @xmath318 as well as @xmath10 and @xmath11 via the data . on the other hand , for each given @xmath319 data matrix @xmath318 , @xmath315 is a universal constant in the threshold @xmath320 in the sense that it does not change with respect to the position @xmath321 .
we also note that the cross - validation is based on the estimate of @xmath25 rather than @xmath0 because poet thresholds the error covariance matrix .
thus cross - validation improves the performance of thresholding .
it is possible to derive the rate of convergence for @xmath322 under the current model setting , but it ought to be much more technically involved than the regular sparse matrix estimation considered by bickel and levina ( 2008 ) and cai and liu ( 2011 ) . to keep our presentation simple
we do not pursue it in the current paper .
we give four examples to which the results in theorems [ thm31][thm33 ] can be applied .
detailed pursuits of these are beyond the scope of the paper .
[ large - scale hypothesis testing ] controlling the false discovery rate in large - scale hypothesis testing based on correlated test statistics is an important and challenging problem in statistics ( leek and storey , 2008 ; efron , 2010 ; fan , et al . , 2012 ) .
suppose that the test statistic for each of the hypothesis @xmath323 is @xmath324 and these test statistics @xmath325 are jointly normal @xmath326 where @xmath80 is unknown .
for a given critical value @xmath327 , the false discovery proportion is then defined as @xmath328 where @xmath329 and @xmath330 are the total number of false discoveries and the total number of discoveries , respectively .
our interest is to estimate @xmath331 for each given @xmath327 .
note that @xmath332 is an observable quantity .
only @xmath333 needs to be estimated .
if the covariance @xmath80 admits the approximate factor structure ( [ eq1.3 ] ) , then the test statistics can be stochastically decomposed as @xmath334 by the principal factor approximation ( theorem 1 , fan , han , gu , 2012 ) @xmath335 when @xmath99 and the number of true significant hypothesis @xmath336 is @xmath337 , where @xmath338 is the upper @xmath327-quantile of the standard normal distribution , @xmath339 and @xmath340 .
now suppose that we have @xmath341 repeated measurements from the model ( [ eq4.1 ] ) .
then , by corollary [ c31 ] , @xmath342 can be uniformly consistently estimated , and hence @xmath343 and @xmath331 can be consistently estimated .
efron ( 2010 ) obtained these repeated test statistics based on the bootstrap sample from the original raw data .
our theory ( theorem [ thm33 ] ) gives a formal justification to the framework of efron ( 2007 , 2010 ) .
the maximum elementwise estimation error @xmath344 appears in risk assessment as in fan , zhang and yu ( 2012 ) . for a fixed portfolio allocation vector @xmath345 , the true portfolio variance and the estimated one
are given by @xmath346 and @xmath347 respectively .
the estimation error is bounded by @xmath348 where @xmath349 , the @xmath45-norm of @xmath345 , is the gross exposure of the portfolio .
usually a constraint is placed on the total percentage of the short positions , in which case we have a restriction @xmath350 for some @xmath351 in particular , @xmath352 corresponds to a portfolio with no - short positions ( all weights are nonnegative ) .
theorem [ thm32 ] quantifies the maximum approximation error .
the above compares the absolute error of perceived risk and true risk .
the relative error is bounded by @xmath353 for any allocation vector @xmath345 .
theorem [ thm32 ] quantifies this relative error .
consider the following panel regression model @xmath354 where @xmath355 is a vector of observable regressors with fixed dimension .
the regression error @xmath356 has a factor structure and is assumed to be independent of @xmath355 , but @xmath6 , @xmath7 and @xmath9 are all unobservable .
we are interested in the common regression coefficients @xmath357 .
the above panel regression model has been considered by many researchers , such as ahn , lee and schmidt ( 2001 ) , pesaran ( 2006 ) , and has broad applications in social sciences .
although ols ( ordinary least squares ) produces a consistent estimator of @xmath357 , a more efficient estimation can be obtained by gls ( generalized least squares ) .
the gls method depends , however , on an estimator of @xmath358 , the inverse of the covariance matrix of @xmath359 . by assuming the covariance matrix of @xmath360 to be sparse
, we can successfully solve this problem by applying theorem [ thm32 ] .
although @xmath356 is unobservable , it can be replaced by the regression residuals @xmath361 , obtained via first regressing @xmath362 on @xmath355 .
we then apply the poet estimator to @xmath363 . by theorem [ thm32 ] ,
the inverse of the resulting estimator is a consistent estimator of @xmath358 under the spectral norm .
a slight difference lies in the fact that when we apply poet , @xmath364 is replaced with @xmath363 , which introduces an additional term @xmath365 in the estimation error .
a celebrated financial economic theory is the capital asset pricing model ( capm , sharpe 1964 ) that makes william sharpe win the nobel prize in economics in 1990 , whose extension is the multi - factor model ( ross , 1976 , chamberlain and rothschild , 1983 ) .
it states that in a frictionless market , the excessive return of any financial asset equals the excessive returns of the risk factors times its factor loadings plus noises . in the multi - period model
, the excess return @xmath2 of firm @xmath3 at time @xmath366 follows model ( [ eq1.1 ] ) , in which @xmath7 is the excess returns of the risk factors at time @xmath366 . to test the null hypothesis ( [ eq1.2 ] ) , one embeds the model into the multivariate linear model @xmath367 and wishes to test @xmath368 .
the f - test statistic involves the estimation of the covariance matrix @xmath25 , whose estimates are degenerate without regularization when @xmath369 .
therefore , in the literature ( sentana , 2009 , and references therein ) , one focuses on the case @xmath10 is relatively small .
the typical choices of parameters are @xmath370 monthly data and the number of assets @xmath371 , 10 or 25 .
however , the capm should hold for all tradeable assets , not just a small fraction of assets . with our regularization technique ,
non - degenerate estimate @xmath222 can be obtained and the f - test or likelihood - ratio test statistics can be employed even when @xmath372 . to provide some insights ,
let @xmath373 be the least - squares estimator of ( [ eq4.3 ] ) . then , when @xmath374 , @xmath375 for a constant @xmath376 which depends on the observed factors .
when @xmath25 is known , the wald test statistic is @xmath377 .
when it is unknown and @xmath10 is large , it is natural to use the f - type of test statistic @xmath378 .
the difference between these two statistics is bounded by @xmath379 since under the null hypothesis @xmath380 , we have @xmath381 .
thus , it follows from boundness of @xmath382 that @xmath383 theorem 3.1 provides the rate of convergence for the above difference .
detailed development is out of the scope of the current paper , and we will leave it as a separate research project .
in this section , we will examine the performance of the poet method in a finite sample .
we will also demonstrate the effect of this estimator on the asset allocation and risk assessment . similarly to fan , et al .
( 2008 , 2011 ) , we simulated from a standard fama - french three - factor model , assuming a sparse error covariance matrix and three factors . throughout this section ,
the time span is fixed at @xmath384 , and the dimensionality @xmath10 increases from @xmath385 to @xmath386 .
we assume that the excess returns of each of @xmath10 stocks over the risk - free interest rate follow the following model : @xmath387 the factor loadings are drawn from a trivariate normal distribution @xmath388 , the idiosyncratic errors from @xmath389 , and the factor returns @xmath7 follow a var(1 ) model . to make the simulation more realistic ,
model parameters are calibrated from the financial returns , as detailed in the following section . to calibrate the model
, we use the data on annualized returns of 100 industrial portfolios from the website of kenneth french , and the data on 3-month treasury bill rates from the crsp database .
these industrial portfolios are formed as the intersection of @xmath390 portfolios based on size ( market equity ) and @xmath390 portfolios based on book equity to market equity ratio .
their excess returns @xmath391 are computed for the period from january @xmath392 , 2009 to december @xmath393 , 2010 . here ,
we present a short outline of the calibration procedure . 1 . given @xmath394 as the input data
, we fit a fama - french - three - factor model and calculate a @xmath395 matrix @xmath396 , and @xmath397 matrix @xmath398 , using the principal components method described in section 3.1 . 2 .
we summarize 100 factor loadings ( the rows of @xmath396 ) by their sample mean vector @xmath399 and sample covariance matrix @xmath400 , which are reported in table 1 .
the factor loadings @xmath401 for @xmath4 are drawn from @xmath402 .
+ .mean and covariance matrix used to generate @xmath403 [ cols="^,^,^,^",options="header " , ]
we study the problem of estimating a high - dimensional covariance matrix with conditional sparsity . realizing unconditional sparsity assumption is inappropriate in many applications , we introduce a latent factor model that has a conditional sparsity feature , and propose the poet estimator to take advantage of the structure .
this expands considerably the scope of the model based on the strict factor model , which assumes independent idiosyncratic noise and is too restrictive in practice . by assuming sparse error covariance matrix , we allow for the presence of the cross - sectional correlation even after taking out the common factors
the sparse covariance is estimated by the adaptive thresholding technique .
it is found that the rates of convergence of the estimators have an extra term approximately @xmath404 in addition to the results based on observable factors by fan et al .
( 2008 , 2011 ) , which arises from the effect of estimating the unobservable factors . as we can see
, this effect vanishes as the dimensionality increases , as more information about the common factors becomes available . when @xmath10 gets large enough , the effect of estimating the unknown factors is negligible , and we estimate the covariance matrices as if we knew the factors .
the proposed poet also has wide applicability in statistical genomics .
for example , carvalho et al . (
2008 ) applied a bayesian sparse factor model to study the breast cancer hormonal pathways .
their real - data results have identified about two common factors that have highly loaded genes ( about half of 250 genes ) . as a result
, these factors should be treated as pervasive " ( see the explanation in example 2.1 ) , which will result in one or two very spiked eigenvalues of the gene expressions covariance matrix .
the poet can be applied to estimate such a covariance matrix and its network model .
* appendix *
we estimate @xmath25 by applying the adaptive thresholding given by ( [ eq2.13 ] ) . however , the task here is slightly different from the standard problem of estimating a sparse covariance matrix in the literature , as no direct observations for @xmath33 are available . in many cases
the original data are contaminated , including any type of estimate of the data when direct observations are not available .
this typically happens when @xmath33 represent the error terms in regression models or when data is subject to measurement of errors .
instead , we may observe @xmath405 .
for instance , in the approximate factor models , @xmath406 we can estimate @xmath25 using the adaptive thresholding proposed by cai and liu ( 2011 ) : for the threshold @xmath407 define @xmath408 @xmath409 where @xmath410 satisfies : for all @xmath149 , @xmath411 @xmath412 when @xmath405 is close enough to @xmath33 , we can show that @xmath413 is also consistent .
the following theorem extends the standard thresholding results in bickel and levina ( 2008 ) and cai and liu ( 2011 ) to the case when no direct observations are available , or the original data are contaminated . for the tail and mixing parameters @xmath414 and @xmath415 defined in assumptions [ a21 ] and [ a32 ] , let @xmath416 .
[ tb1 ] suppose @xmath417 , and assumptions [ a21 ] and [ a32 ] hold .
in addition , suppose there is a sequence @xmath418 so that @xmath419 and @xmath420 then there is a constant @xmath146 in the adaptive thresholding estimator ( [ eq3.2add ] ) with @xmath421 such that @xmath422 if further @xmath423 , then @xmath424 is invertible with probability approaching one , and @xmath425 by assumptions [ a21 ] and [ a32 ] , the conditions of lemmas a.3 and a.4 of fan , liao and mincheva ( 2011 , _ ann .
, * 39 * , 3320 - 3356 ) are satisfied .
hence for any @xmath426 , there are positive constants @xmath427 and @xmath428 such that each of the events @xmath429 occurs with probability at least @xmath430 . by the condition of threshold function , @xmath431 .
now for @xmath432 under the event @xmath433 @xmath434 let @xmath435 then with probability at least @xmath436 , @xmath437 since @xmath296 is arbitrary , we have @xmath438 .
if in addition , @xmath423 , then the minimum eigenvalue of @xmath424 is bounded away from zero with probability approaching one since @xmath184 .
this then implies @xmath439
we first cite two useful theorems , which are needed to prove propositions 2.1 and 2.2 . in lemma [ lb.1 ] below , let @xmath440 be the eigenvalues of @xmath80 in descending order and @xmath441 be their associated eigenvectors .
correspondingly , let @xmath442 be the eigenvalues of @xmath443 in descending order and @xmath102 be their associated eigenvectors .
[ lb.1 ] 1 . *
( weyl s theorem ) * @xmath444 .
( @xmath86 theorem , davis and kahan , 1970 ) * @xmath445 * proof of proposition 2.1 * since @xmath79 are the eigenvalue of @xmath0 and @xmath78 are the first @xmath12 eigenvalues of @xmath72 ( the remaining @xmath446 eigenvalues are zero ) , then by the weyl s theorem , for each @xmath84 , @xmath447 for @xmath448 , @xmath449 on the other hand , the first @xmath12 eigenvalues of @xmath450 are also the eigenvalues of @xmath77 . by the assumption , the eigenvalues of @xmath75 are bounded away from zero .
thus when @xmath84 , @xmath451 are bounded away from zero for all large @xmath452 * proof of proposition 2.2 * applying the @xmath453 theorem yields @xmath454 for a generic constant @xmath61 , @xmath455 for all large @xmath10 , since @xmath456 but @xmath457 is bounded by prosposition 2.1 . on the other hand , if @xmath458 , the same argument implies @xmath459 . if @xmath460 , @xmath461 , where @xmath462 is bounded away from zero , but @xmath463 .
hence again , @xmath464 * proof of theorem [ thm2.1 ] * the sample covariance matrix of the residuals using least squares method is given by @xmath465 where we used the normalization condition @xmath466 and @xmath467 if we show that @xmath468 , then from the decompositions of the sample covariance @xmath469 we have @xmath470 .
consequently , applying thresholding on @xmath471 is equivalent to applying thresholding on @xmath472 , which gives the desired result .
we now show @xmath468 indeed holds .
consider again the least squares problem ( [ eq2.10 ] ) but with the following alternative normalization constraints : @xmath473 and @xmath474 is diagonal .
let @xmath475 be the solution to the new optimization problem .
switching the roles of @xmath55 and @xmath131 , then the solution of ( [ eq2.12 ] ) is @xmath476 and @xmath477 .
in addition , @xmath478 from @xmath479 , it follows that @xmath480
we will proceed by subsequently showing theorems [ thm33 ] , [ thm31 ] and [ thm32 ] . [
la2 ] suppose that the random variables @xmath487 both satisfy the exponential - type tail condition : there exist @xmath414 , @xmath488 and @xmath188 , such that @xmath489 , @xmath490 then for some @xmath415 and @xmath491 , and any @xmath189 , @xmath492 first of all , by proposition [ prop21 ] , under assumption [ a35 ] , the @xmath500 largest eigenvalue @xmath501 of @xmath0 satisfies : for some @xmath502 @xmath503 for sufficiently large @xmath10 . using weyl s theorem , we need only to prove that @xmath504 .
without loss of generality , we prove the result under the identifiability condition ( [ eq2.7 ] ) . using model ( [ eq1.2 ] ) ,
@xmath505 using this and ( [ eq1.3 ] ) , @xmath506 can be decomposed as the sum of the four terms : @xmath507 we now deal them term by term .
we will repeatedly use the fact that for a @xmath18 matrix @xmath40 , @xmath508 first of all , by lemma [ lb4 ] , @xmath509 which is @xmath510 if @xmath511 .
consequently , by assumption [ a35 ] , we have @xmath512 we now deal with @xmath513 .
it follows from lemma [ lb4 ] that @xmath514 since @xmath515 , it remains to deal with @xmath516 , which is bounded by @xmath517 which is @xmath510 since @xmath518 .
since @xmath33 is weakly stationary , @xmath520 in addition , @xmath521 for some constant @xmath165 and any @xmath522 since @xmath9 has exponential tail .
hence by davydov s inequality ( corollary 16.2.4 in athreya and lahiri 2006 ) , there is a constant @xmath146 , for all @xmath523 , @xmath524 , where @xmath525 is the @xmath526-mixing coefficient . by assumption [ a32 ]
, @xmath527 thus uniformly in @xmath11 , @xmath528 our derivation below relies on a result obtained by bai and ng ( 2002 ) , which showed that the estimated number of factors is consistent , in the sense that @xmath163 equals the true @xmath36 with probability approaching one . note that under our assumptions [ a35]-[a33 ] , all the assumptions in bai and ng ( 2002 ) are satisfied .
thus immediately we have the following lemma .
\(i ) when @xmath575 , by lemma [ lb.5 ] , all the eigenvalues of @xmath576 are bounded away from zero . using the inequality @xmath577 and the identity ( [ eb1 ] ) , we have , for some constant @xmath146 , @xmath578 each of the four terms on the right hand side above are bounded in lemma [ lb1 ] , which then yields the desired result .
we first condition on @xmath575 .
( i ) lemma [ lb.5 ] implies @xmath582 .
also @xmath583 in addition , @xmath584 .
it then follows from the definition of @xmath263 that @xmath585 .
define @xmath586 applying the triangular inequality gives : @xmath587 by lemma [ lb4 ] , the first term in ( [ ec.6 ] ) is @xmath588 the second term of ( [ ec.6 ] ) can be bounded , by the cauchy - schwarz inequality and lemma [ lb3 ] , as follows : @xmath589 ( ii ) still conditioning on @xmath575 , since @xmath590 and @xmath585 , right multiplying @xmath263 gives @xmath591 . part
( i ) also gives , conditioning on @xmath575 , @xmath592 .
hence further left multiplying @xmath593 yields @xmath594 .
due to @xmath595 , we reach the desired result . using the facts that @xmath597 , and that @xmath598 , we have @xmath599 we bound the three terms on the right hand side respectively .
it follows from lemmas [ lb4 ] and [ lb5 ] that @xmath600 for the second term , @xmath601
. therefore , @xmath602 .
the cauchy - schwarz inequality and lemma [ lb3 ] imply @xmath603 finally , @xmath604 and @xmath605 imply that the third term is @xmath606 we have , @xmath611 therefore , using the inequality @xmath612 , we have : @xmath613 the first part of the lemma then follows from theorem [ thm33 ] and lemma [ lb3 ] .
the second part follows from corollary [ c31 ] .
\(i ) we have @xmath620 moreover , since all the eigenvalues of @xmath0 are bounded away from zero , for any matrix @xmath40 , @xmath621 .
hence @xmath622 + ( ii ) by theorem [ thm31 ] , @xmath623 + ( iii ) the same argument of the proof of theorem 2 in fan , fan and lv ( 2008 ) implies that @xmath624 .
thus , @xmath625 is upper bounded by @xmath626 + ( iv ) again , by @xmath624 , and lemma [ lb5 ] , @xmath627 \(i ) by lemma [ lb5 ] , with probability approaching one , @xmath637 is bounded away from zero .
hence , @xmath638 ( ii ) the result follows from part ( i ) and lemma [ lc.13 ] .
part ( iii ) and ( iv ) follow from a similar argument of part ( i ) and lemma [ lb5 ] .
we derive the rate for @xmath639 define @xmath640 note that @xmath641 and @xmath642 the triangular inequality gives @xmath643 using the sherman - morrison - woodbury formula , we have @xmath644 where @xmath645^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}'({\widehat{\mathbf{\sigma}}}_{u , \widehat{k}}^{\mathcal{t}})^{-1}\| \cr l_3&=&\|(({\widehat{\mathbf{\sigma}}}_{u , \widehat{k}}^{\mathcal{t}})^{-1}-{\mathbf{\sigma}}_u^{-1}){\widehat { \mbox{\boldmath $ \lambda$}}}[{\mathrm{\bf i}}_k+{\widehat { \mbox{\boldmath $ \lambda$}}}'({\widehat{\mathbf{\sigma}}}_{u , \widehat{k}}^{\mathcal{t}})^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}]^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}'{\mathbf{\sigma}}_u^{-1}\| \cr l_4&=&\|{\mathbf{\sigma}}_u^{-1}({\widehat { \mbox{\boldmath $ \lambda$}}}-{{\mathrm{\bf b}}}{\mathrm{\bf h}}')[{\mathrm{\bf i}}_k+{\widehat { \mbox{\boldmath $ \lambda$}}}'({\widehat{\mathbf{\sigma}}}_{u , \widehat{k}}^{\mathcal{t}})^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}]^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}'{\mathbf{\sigma}}_u^{-1}\| \cr l_5&=&\|{\mathbf{\sigma}}_u^{-1}({\widehat { \mbox{\boldmath $ \lambda$}}}-{{\mathrm{\bf b}}}{\mathrm{\bf h}}')[{\mathrm{\bf i}}_k+{\widehat { \mbox{\boldmath $ \lambda$}}}'({\widehat{\mathbf{\sigma}}}_{u , \widehat{k}}^{\mathcal{t}})^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}]^{-1}{\mathrm{\bf h}}{{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}\| \cr l_6&=&\|{\mathbf{\sigma}}_u^{-1}{{\mathrm{\bf b}}}{\mathrm{\bf h}}'([{\mathrm{\bf i}}_k+{\widehat { \mbox{\boldmath $ \lambda$}}}'({\widehat{\mathbf{\sigma}}}_{u , \widehat{k}}^{\mathcal{t}})^{-1}{\widehat { \mbox{\boldmath $ \lambda$}}}]^{-1}-[{\mathrm{\bf i}}_k+{\mathrm{\bf h}}{{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}{{\mathrm{\bf b}}}{\mathrm{\bf h}}']^{-1}){\mathrm{\bf h}}{{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}\|.\end{aligned}\ ] ] we bound each of the six terms respectively .
first of all , @xmath45 is bounded by theorem [ thm31 ] .
let @xmath646^{-1}$ ] , then @xmath647 note that theorem [ thm31 ] implies @xmath648 .
lemma [ lc.14 ] then implies @xmath649 this shows that @xmath650 .
similarly @xmath651 .
in addition , since @xmath652 , @xmath653 similarly @xmath654 . finally , let @xmath655^{-1}.$ ] by lemma [ lc.14 ] , @xmath656 .
then by lemma [ lc.13 ] , @xmath657 consequently , @xmath658 adding up @xmath45-@xmath659 gives @xmath660 one the other hand , using sherman - morrison - woodbury formula again implies @xmath661^{-1}-[{\mathrm{\bf i}}_k+{{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}{{\mathrm{\bf b}}}]^{-1}){{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}\|\cr & \leq & o(p)\|[({\mathrm{\bf h}}'{\mathrm{\bf h}})^{-1}+{{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}{{\mathrm{\bf b}}}]^{-1}-[{\mathrm{\bf i}}_k+{{\mathrm{\bf b}}}'{\mathbf{\sigma}}_u^{-1}{{\mathrm{\bf b}}}]^{-1}\|\cr & = & o_p(p^{-1})\|({\mathrm{\bf h}}'{\mathrm{\bf h}})^{-1}-{\mathrm{\bf i}}_k\|=o_p(\omega_t^{1-q}m_p).\end{aligned}\ ] ] we first bound @xmath663 . repeatedly using the triangular inequality
yields @xmath664\cr & \leq&(\max_i\|{\widehat { \mathrm{\bf b}}}_i-{\mathrm{\bf h}}{\mathrm{\bf b}}_i\|)^2 + 2\max_{ij}\|{\widehat { \mathrm{\bf b}}}_i-{\mathrm{\bf h}}{\mathrm{\bf b}}_i\|\|{\mathrm{\bf h}}{\mathrm{\bf b}}_j\|+\max_i\|{\mathrm{\bf b}}_i\|^2\|{\mathrm{\bf h}}'{\mathrm{\bf h}}-{\mathrm{\bf i}}_k\|\cr & = & o_p(\omega_t).\end{aligned}\ ] ] on the other hand , let @xmath665 be the @xmath321 entry of @xmath25
. then @xmath666 .
@xmath667 hence @xmath668 the result then follows immediately .
carvalho , c. , chang , j. , lucas , j. , nevins , j. , wang , q. and west , m. ( 2008 ) .
high - dimensional sparse factor modeling : applications in gene expression genomics . _ j. amer .
assoc . _ * 103 * , 1438 - 1456 .
johnstone , i. m. ( 2001 ) . on the distribution of the largest eigenvalue in principal components analysis .
_ , * 29 * , 295327 .
johnstone , i.m . and
( 2009 ) . on consistency and sparsity for principal components analysis in high dimensions .
_ j. amer .
assoc . _ * 104 * , 682 - 693 .
witten , d.m .
, tibshirani , r. and hastie , t. ( 2009 ) . a penalized matrix decomposition , with applications to sparse principal components and canonical correlation analysis . _ biostatistics _ , * 10 * , 515 - 534 .
wright , j. , peng , y. , m , y. , ganesh , a. and rao , s. ( 2009 ) .
robust principal component analysis : exact recovery of corrupted low - rank matrices by convex optimization .
_ manuscript_. microsoft research asia | this paper deals with the estimation of a high - dimensional covariance with a conditional sparsity structure and fast - diverging eigenvalues . by assuming sparse error covariance matrix in an approximate factor model , we allow for the presence of some cross - sectional correlation even after taking out common but unobservable factors . we introduce the principal orthogonal complement thresholding ( poet ) method to explore such an approximate factor structure with sparsity .
the poet estimator includes the sample covariance matrix , the factor - based covariance matrix ( fan , fan , and lv , 2008 ) , the thresholding estimator ( bickel and levina , 2008 ) and the adaptive thresholding estimator ( cai and liu , 2011 ) as specific examples .
we provide mathematical insights when the factor analysis is approximately the same as the principal component analysis for high - dimensional data .
the rates of convergence of the sparse residual covariance matrix and the conditional sparse covariance matrix are studied under various norms .
it is shown that the impact of estimating the unknown factors vanishes as the dimensionality increases .
the uniform rates of convergence for the unobserved factors and their factor loadings are derived .
the asymptotic results are also verified by extensive simulation studies .
finally , a real data application on portfolio allocation is presented .
* keywords : * high - dimensionality , approximate factor model , unknown factors , principal components , sparse matrix , low - rank matrix , thresholding , cross - sectional correlation , diverging eigenvalues . |
the recent discovery of superconductivity below t@xmath4 in na@xmath5co0@xmath6h@xmath2o@xmath7 @xcitehas aroused much research interest .
this material constitutes another example of superconducting transition metal oxides with layered structure besides the cuprates .
the similarity between the two systems make it possible that the study of the new superconductor may shed light on those difficult problems in the study of the cuprates . in na@xmath1coo@xmath2 material ,
the co@xmath8 ion possess a half spin and play a similar role as the cu@xmath9 ion in cuprates .
however , the co@xmath8 in na@xmath1coo@xmath2 form a triangular lattice rather than square lattice as in the cuprates . from theoretical point of view , triangular lattice system is even more interesting since it is intrinsically frustrated . shortly after the discovery of superconductivity
several theoretical proposals are raised to address its mechanism@xcite . on the basis of the mean field analysis on @xmath0 model , a @xmath10 pairing state
is proposed by barskran@xcite , while tanaka et al proposed a triplet pairing state based on symmetry consideration@xcite . experimentally , the pairing symmetry is still under debate@xcite . besides superconductivity ,
this system also exhibit nontrivial magnetic properties .
in particular , a phase transition at 22 @xmath11 from paramagnetic to weak ferromagnetic state is reported in na@xmath12coo@xmath2@xcite .
experience in the study of cuprates tell us that the magnetic degree of freedom may play an important role in the low energy physics of this strongly correlated system . especially , it is important to find out the relation between the magnetic properties and the superconductivity .
thus a comprehensive understanding of the magnetic properties of this system is valuable . in this respect ,
singh calculated the magnetic properties of the system in the lsda scheme@xcite and found that the carrier in the conduction band is always fully polarized , even in the half filled case .
this result is inconsistent with the result of the strong coupling analysis based on @xmath0 model . in the lsda - like weak coupling treatment ,
the tendency of the system toward ferromagnet state is exaggerated due to the absence of electron correlation(the electron avoid each other only through the pauli exclusion principle which make the ferromagnetic ordering of the electron spin energetically favorable ) . in technical terms , the stoner factor @xmath13 overestimate the ferromagnetic instability in the strongly correlated regime .
in fact , as @xmath14 increase , the density of state at the fermi energy is also depressed .
especially , at the half filled case , @xmath15 may reduce to zero due to the mott insulator physics .
thus a through study of the magnetic properties of the system in the strong coupling regime is deserved . in this paper , we study the magnetic properties of the system within the@xmath0 model perspective .
complementary to the weak coupling picture , in the strong coupling regime , it is the reduction of kinetic energy rather than the potential energy that dominate the low energy physics and thus determine the magnetic properties since the no double occupancy constraints in the @xmath0 model already take into account of the effect of the potential energy in the zeroth order . here
the kinetic energy contains contribution from both real charge transport of doped holes and the virtual hopping process that lead to spin exchange .
the real and virtual part of the kinetic energy are competing with each other . the real kinetic energy favor ferromagnetic alignment of the spin due to the nagaoka physics while the virtual kinetic energy or the superexchange favor antiferromagnetic alignment of spin since it is blocked by pauli exclusion principle .
at half filling , the real process is depressed and the virtual process dominate .
thus a fully polarized state is obviously unfavorable in this case .
in fact , the triangular @xmath0 model show zero net magnetization at the half filling and is generally believed to be in a coplanar 120@xmath16 state with three sublattices@xcite(see figure 1 ) .
how dose this state evolve with doping ?
this is the central problem we want to address in this paper .
we find the answer depend crucially on the sign of the hoping term of the @xmath0 model(on a triangular lattice , the particle hole symmetry is broken and the sign of the hopping term is essential ) .
a negative hopping matrix element will induce a @xmath17 -flux around each triangular plaqutte of the lattice and frustrate the motion of the charge carrier .
the charge carrier will also see another flux due to the spin berry phase in a noncollinear spin background .
the interplay between the two result in a complex doping dependence of the magnetic properties found in this work .
although the hopping term in the na@xmath1coo@xmath2 system is positive , we will consider both positive and negative sign case in our paper for general interest . in this paper
, we treat the @xmath0 model in the slave fermion - schwinger boson formalism since it is convenient for the discussion of magnetic properties . at half filling , the spin system is in a coplanar state with three sublattices .
the spin in the three sublattices lay symmetrically on their common plane with a 120@xmath16 angle between each other . from this state
, the most natural way for the system to obtain a net magnetization is to tilt the spin of the three sublattices symmetrically above their common plane .
this is supported by a unrestricted search of the semiclassical energy functional .
the state so obtained is called a canting state . with this picture in mind
, we propose a sound ansatz for the mean field rvb order parameter for the spins .
we call the mean field ansatz constructed in this way the canting ansatz . working with the canting ansatz , we map out the whole doping - temperature phase diagram of the @xmath0 model for both sign of the hopping term .
the main result result of this paper can be summarized as follow .
we find there is large doping range in which the system show zero net magnetization , rather than saturated magnetization as predicted in the lsda calculation .
thus the lsda result is unreliable in the strong coupling regime .
we find the spin berry phase play a vital role in this geometrically frustrated system .
we find the various states in the phase diagram are characterized(and distinguished ) by their respective spin berry phase , rather than any landau - like order parameter related to broken symmetry .
we also find the spin berry phase is responsible for the qualitative difference in the low energy excitation spectrum of the various states in the phase diagram .
we argue that the phase boundary in the mean field phase diagram are true phase transition lines and may serve as the first explicit and realistic example for phase transition between states with different quantum orders@xcite .
we argue the spin berry phase provide a first realistic example for the concept of quantum order .
we find there exists an exotic state with nonzero spin chirality but no spin ordering in a large temperature and doping range .
we find this exotic state support nonzero staggered current loop in the bulk of the system and propose to study this exotic state in the hole doped na@xmath3tio@xmath2 system . as a by - product of this study
, we also achieve an improvement over earlier schwinger boson mean field theory on the triangular antiferromagnet .
the canting ansatz proposed in this paper lead to two branches of gapless spinon modes .
we find the small gap of the third spinon mode is caused by quantum fluctuation and our theory reproduce the linear spin wave theory result in the semiclassical limit .
finally , we find our general phase diagram is consistent with experimental result on na@xmath3coo@xmath2 system and a singlet pairing picture may be more relevant in the small doping regime .
the paper is organized as follows . in section
ii , the schwinger boson - slave fermion representation for the @xmath0 model and our working ansatz , the canting ansatz are introduced .
the phase diagram is presented in section iii . in section iv ,
we present the result for the doping dependence of magnetization and excitation spectrum . in the concluding section iv , relevance of our result to experiments
are discussed .
our starting point is the @xmath0 model on triangular lattice , @xmath18 in which @xmath19 satisfy the no double occupation constraint @xmath20 .
@xmath21 denotes sum over nearest - neighbouring sites . for the triangular lattice ,
the sign the @xmath22 is essential and we will consider both sign of @xmath22 for theoretical interest , although @xmath22 is positive in na@xmath1coo@xmath2 system . at half filling , the model reduce to the much studied antiferromagnetic heisenberg model on a triangular lattice .
it is believed that this model posses a coplanar magnetic order with three sublattices at zero temperature@xcite . in this state ,
the spins in the three sublattices reside symmetrically on their common plane with 120@xmath16 angle between each other(see figure 1 ) .
this result can be obtained in the schwinger - boson mean field theory of the model@xcite . in this formalism
, the spin operator @xmath23 is represented by schwinger boson @xmath24 @xmath25 with the constraint @xmath26 and the hamiltonian is @xmath27 in the mean field treatment , order parameter @xmath28 , @xmath29 are introduced to represent the spin correlation in the system .
although @xmath30 and @xmath31 are themselves spin rotational invariant , by condensing the bosonic spinon @xmath32 we can still break this symmetry and describe magnetic ordered state(note both @xmath30 and @xmath31 are needed for a nonbipartite lattice to find the absolute minima of the energy functional ) .
the choice of the mean field ansatz @xmath30 and @xmath31 depends on our understanding of the ground state .
the ansatz used in previous theories@xcite on this problem can reproduce the three sublattice magnetic order , but fails to give three branches of gapless spin wave and are thus not fully satisfactory . in the following
, we will determine the mean field ansatz with a semiclassical analysis .
when the system is doped , we should introduce another slave particle to represent the hole degree of freedom . in this paper , we use the slave fermion- schwinger boson representation . in this formalism
@xmath33 in which @xmath34 is fermionic operator for the holon , and @xmath35 is bosonic operator for the spinon just defined . the no double occupation constraint is now expressed as : @xmath36and the hamiltonian ( [ 1 ] ) is : @xmath37 in which @xmath38 is the lagrangian multipliers to enforce particle number constraint and @xmath39 is the chemical potential for the hole .
this hamiltonian can be decoupled into bilinear form with the introduction of the mean field order parameter @xmath30 and @xmath31 . in principle
, we should also include the order parameter @xmath40 in the doped case .
however , since the hole has no dynamics of its own , @xmath41 is proportional to @xmath42 and is not an independent variable .
after the decoupling , we get the mean field hamiltonian @xmath43 -\mu _ { b}\dsum\limits_{i}b_{i\alpha } ^{\dagger } b_{i\alpha } ^{{}}\ ] ] for spinon and @xmath44 -\mu _ { f}\dsum\limits_{i}f_{i}^{\dagger } f_{i}\]]for holon . here
@xmath45 is the antisymmetric tensor . @xmath46 and @xmath47 are the spinon and holon chemical potential to enforce the particle number constraints @xmath48 and @xmath49 , where @xmath50 is the hole concentration . from this hamiltonian
, we see the spin background affect the hole motion through the order parameter @xmath31 whose phase is nothing but the spin berry phase mentioned in the introduction .
now we determine the mean field ansatz @xmath30 and @xmath31 from a semiclassical analysis@xcite . in the semiclassical limit
, we can take @xmath32 as a two component spinor @xmath51 in which @xmath52 and @xmath53 are spherical coordinates of the spin @xmath54 . here
an arbitrary gauge phase factor @xmath55 is omitted .
now let us consider possible magnetization of the system from the coplanar state at half filling .
in this parent state , the spins in the three sublattices lay symmetrically in their common plane .
hence the most natural way for the system to obtain a net magnetization is to tilt the spins of the three sublattices symmetrically above their common plane .
we call such a state a canting state . in this state ,
the semiclassical spin on the three sublattices are@xmath56where @xmath57 stands for the three sublattices .
the coplanar state at half filling is a special case of the canting state with @xmath58 .
this canting ansatz is supported by minimizing the semiclassical energy functional .
this energy functional is obtained by substituting ( [ 3 ] ) into ( [ 2])(with @xmath59 replaced by @xmath60 in the doped case , where @xmath61 is the hole concentration ) and then diagonalize the holon hamiltonian .
we performed an unconstraint search on a @xmath62 lattice and find the canting state always has the lowest energy . in the semiclassical limit , the mean field order parameter can be readily written down . for the canting state ,
we have @xmath63 to fix the gauge totally , we require @xmath64 to be real . in this paper , we will study mean field ansatz satisfying ( [ 4 ] ) and call such a mean field ansatz canting ansatz . in this ansatz , a real order parameter @xmath64 and a complex order parameter @xmath65
are introduced to represent the spin correlation in the system .
in particular , the phase of @xmath65 equals to one sixth of the solid angle spanned by the spins in the three sublattices and is thus related to the spin chirality and the spin berry phase of the background spins@xcite
. the mean field hamiltonian with the canting ansatz can be diagonalized . introducing @xmath66 and @xmath67
, the hamiltonian can be written as @xmath68 where the expression for the @xmath69 matrix @xmath70 and the @xmath71 matrix @xmath72 are given in the appendix . here
the momentum sum runs over the reduced brillouin zone for the three sublattice system .
the six branches of spinon dispersions are @xmath73 ^{2}-\frac{1}{4}\left [ d\func{im}(\gamma _ { n , k}^{{}})% \right ] ^{2}}\right\vert\ ] ] with @xmath74 .
the three branches of holon dispersions are @xmath75 with @xmath74 . here
@xmath76 , @xmath77 and @xmath78 in which @xmath79 since @xmath80 , each branch of spinon mode is two - fold degenerate .
this is a reflection of the spin rotational symmetry which is unbroken in our treatment .
by solving the self - consistent mean field equations , we can obtain a series of solutions for @xmath64 and @xmath65 .
these solutions are characterized by the phase of @xmath65 .
as we have shown , this phase angle , @xmath81 , equals to one sixth of the solid angle spanned by the spins .
since the solid angle is defined modulo @xmath82 and the problem is symmetric under the reflection about the common plane of the spins , we can restrict ourselves to the interval @xmath83 . in the following
, the state with @xmath84 will be called a collinear state and the state with @xmath85 will be called the coplanar state . a state with @xmath81 between this two limits
will be called the canted state . in the next section
, we will map out the mean field phase diagram of the model by determining the solution with the lowest free energy .
the temperature - doping phase diagram for @xmath87 is shown in figure 2 . at low doping ,
the coplanar state has the lowest free energy . above a critical doping about @xmath88
the system transform into the collinear state through a first order phase transition .
the canted state is never stable for @xmath86 .
this result can be understood as follows . from ( [ 1]),we
see the bare hopping term of the hole ( which is @xmath89 ) is renormalized by the spin dependent factor @xmath65 . for @xmath86 ,
the hole will feel a flux of strength @xmath90 around each elementary triangular loop of the lattice(@xmath81 is the phase of @xmath65 ) and is in general frustrated . when @xmath91 , or , when the spin is in the coplanar state , the frustration due to the bare hopping term is released by the spin berry phase from @xmath65 .
thus at low doping , the spins prefer to stay in the coplanar state(this state is of course also favored by the spin exchange energy ) . at high doping ,
the situation is reversed and now it is the holes of the nearly filled hole band that are moving .
the sign of the hopping term of these holes is opposite to that of the original hole and thus the frustration due the bare hopping term is absent .
thus at high doping , the spins prefer to align themselves ferromagnetically(the spin exchange energy is suppressed at high doping due to dilution of spins ) . in the argument raised above , we neglect the effect of the amplitude of the @xmath65 .
semiclassical analysis shows that this neglect has no qualitative effect . at zero temperature
, the condensation of the spinon will lead to long range order of spins . in the coplanar state , there are two gapless spinon modes(see discussion in section iv ) .
both modes condense into the coplanar spin long range order . in the collinear state , there is only one gapless spinon mode .
the condensation of this mode give rise to ferromagnetic long range order of the spins .
thus unlike the lsda result , there is a finite doping range in our phase diagram in which the net magnetization is zero . at finite temperature ,
the spin long range order does not exist .
the distinction between the coplanar state and the collinear state is now more subtle . on symmetry ground
, it is hard to tell apart between the two . in our theory
, the two states are distinguished by their spin berry phase .
the problem here is whether the phase boundary between the coplanar state and the collinear state represent true phase transition or just crossover .
we think this phase boundary still represent true phase transition since there is a discontinues change of the spin correlation pattern(embodied in the spin berry phase ) across it .
this discontinues change in the spin correlation pattern lead to a discontinues change in the excitation spectrum and thus thermodynamical properties of the system across the phase boundary .
as will be shown in the next section , the spin berry phase is responsible for the qualitative difference in the excitation spectrum of the various states in the phase diagram . in particular , in the coplanar state with @xmath91 , there is always two branches of gapless spinon(the gap due to finite temperature is exponentially small ) , both of which have linear dispersion .
while in the collinear state ( or any states with @xmath92 ) , there is only one branch of gapless spinon whose dispersion is quadratic . in other words , the two branches of gapless spinon mode in the coplanar state seems to be protected by some kind of order not related to any broken symmetry .
this special kind of the order , which is nothing but the characteristic spin berry phase of the coplanar state , is broken across the phase boundary between the coplanar state and the collinear state .
it is in this sense that we think the phase boundary a true phase transition line .
the concept of an order with no broken symmetry is first advocated by wen who give it the name quantum order@xcite .
the quantum order , as it defined by wen , describes the structure in the quantum wave function which is beyond the classification on symmetry ground and can be detected by checking the excitation spectrum of the system which is sensitive to the structure of the ground state wave function . in our case , the quantum order is nothing but the spin berry phase .
we think the transition between the coplanar state and the collinear state may serve as the first explicit and realistic example for phase transition between states with different quantum orders .
this is the most important finding of this paper . in this regard ,
it is seems natural to look at the zero temperature transition between the two states also as a transition between different quantum orders rather than the conventional landau - type phase transition , although the symmetry is really broken in this case .
the realistic value of @xmath93 for na@xmath1coo@xmath2 is still unsettled .
in our theory , the phase boundary between the coplanar state and the fully polarized state will shift to lower doping with increasing @xmath93 .
if our assignment of @xmath87 is realistic(arpes experiment report a similar value@xcite ) , then the superconductivity observed around @xmath94 is covered totally in the coplanar state in which ferromagnetic fluctuation is suppressed .
this implies that the singlet pairing picture may be more relevant for the superconductivity in this system .
furthermore , the ferromagnetic transition observed at @xmath95 for @xmath96 @xciteseems to be also consistent with our phase diagram .
the temperature - doping phase diagram for @xmath98 is shown in fig.3 .
the phase diagram is much more complex as compared with that of the @xmath86 case . at zero temperature ,
the spins ca nt gradually out of their common plane with increasing doping until a critical doping level , above which the spin return back into the coplanar state through a first order phase transition . at finite temperature ,
the phase diagram show complex structures . below a lower critical doping , the zero temperature canted state return back into the coplanar state through a second order phase transition at high temperature . above the lower critical doping , the canted state transform into the collinear state through a second order phase transition .
the coplanar state at high doping also transform into the collinear state at high temperature , but through a first order phase transition .
the phase diagram contains both first order and second order phase transition lines with tricritical and quadricritical points join them .
this complex phase diagram structure can also be understood on semiclassical ground . for @xmath97 , the frustration due to the bare hopping term is absent and the kinetic energy of the hole
favor a state with @xmath84 , or , the collinear state at low doping .
the antiferromagnetic exchange energy act against this tendency .
the compromise between the two result in the canted state in which the spin ca nt gradually out of the coplanar state at half filling . at high doping ,
the situation is again reversed and now the kinetic energy favors the coplanar state which is also favored by the exchange energy . below the lower critical doping , the spin canting at zero temperature induced by the hole will eventually be erased by the thermal fluctuation and the spins will return to the coplanar state at high temperature . at higher doping level
, the hole system is more robust and it in fact help to enhance the spin correlation against thermal fluctuation .
this is quite different from the situation in cuprates . in the triangular lattice
, the spin is already frustrated at half filling . doping a small concentration of holes in fact help to release such geometrical frustration .
this special effect will be discussed further in the next section . at finite temperature ,
the spin long range order can not exist and the various phases in the phase diagram are characterized by the value of the spin berry phase . among these phases , of particular interest is the canted phase which hold a special kind of long range order even at finite temperature .
in the canted phase , the time reversal symmetry is spontaneously broken and system posses a nonzero spin chirality .
thus the phase boundary between the canted phase and other phases are true phase transitions rather than crossovers even in the landau sense . in the canted phase ,
the spin chirality will induce a nonzero current of the hole .
since the spin chirality is staggered , the induced current is also staggered .
thus in the canted state , we expect staggered current to flow around each triangular loop .
this is the most important characteristic of the canted phase .
a detection of such current loop is interesting . before closing this subsection
we note that although the @xmath97 case of the model is not directly relevant to the na@xmath3coo@xmath2 system , it is proposed that it may be appropriate for the hole doped system na@xmath3tio@xmath99@xcite .
at zero temperature , the gapless spinon will condense into magnetic long range order . for the canted state and the collinear state , this condensation will lead to a nonzero net magnetization of the system . for
the @xmath86 case(which is relevant for the na@xmath3coo@xmath2 system ) , there is discontinues jump of the magnetization at the transition point between the coplanar state and the collinear state . for @xmath97 , the magnetization build up gradually with increasing doping as the spin ca nt out of their common plane and jump to zero when the spin return back into the coplanar state at high doping .
these results are shown in figure 4 .
we also presents the result for the fraction of the condensed spin ( which is a measure of magnitude of sublattice magnetization)in this figure .
we see doping a small amount of hole enhance the sublattice magnetization for both @xmath86 and @xmath97 .
the reason for this enhancement is that the doped hole can release partially the geometric frustration inherent of the triangular lattice . in quantum systems ,
the structures in the ground state wave function often have nontrivial consequences in the excitation spectrum of the system .
in fact , the excitation spectrum serve as the most direct way to detect any non - symmetry - breaking related structure(or quantum order ) in the ground state wave function of quantum system . in our case
, the various phases in the phase diagram are distinguished by their spin berry phase .
now we discuss the effect of the spin berry phase on the excitation spectrum of the system . in the reduced brillouin zone of the three sublattice system , there are three branches of nondegenerate spinon modes .
these three branches of nondegenerate spinon modes transform into each other under the translation in the momentum space by a reciprocal lattice vector of the reduced brillouin zone(this can be easily shown since @xmath100 , where @xmath101 is a reciprocal lattice vector of reduced brillouin zone ) .
thus in the extended zone scheme , these three branches of spinon modes can be expressed by a single formula @xmath102 ^{2}-\frac{1}{4}\left [ d% \func{im}(\gamma _ { k})\right ] ^{2}}\right\vert \label{5}\ ] ] in the coplanar state @xmath103 is real(since @xmath85 ) and the spinon spectrum can be further reduced to @xmath104 ^{2}-\frac{1% } { 4}\left [ d\func{im}(\gamma _ { k})\right ] ^{2}}\ ] ] this spectrum is degenerate at the six corners of the brillouin zone of the triangular lattice(@xmath105at these points ) .
this degeneracy is the most important characteristic of the excitation spectrum of the coplanar state and it is protected by the spin berry in this state . when @xmath81 deviate from @xmath106 , this degeneracy
is gone . at zero temperature ,
the spinon at these momentums will become gapless and condense into the @xmath107 spin long range order .
thus , at general doping there are two branches of gapless spin wave excitation on the ordered spin background(the six corners of the brillouin zone correspond to two independent momentums ) . since @xmath108 for small @xmath109 , both of the two modes have linear dispersion around the gapless point .
besides these two gapless modes , there is a third minimum of the excitation spectrum at the center of the brillouin zone .
the energy of this mode decrease linearly with decreasing doping but remain finite at half filling as shown figure 5 .
the excitation spectrum of the half filled system is shown in figure 6 and it is compared with spin wave spectrum obtained in the linear spin wave theory(lswt)@xcite .
the lswt predict three branches of gapless spin wave on the three sublattice spin background while our schwinger boson mean theory predict only two .
the reason for this discrepancy can be seen in the following way . requiring both the zone corner modes and the zone center to be gapless
, we obtain @xmath110 at half filling , a relation valid for coplanar state in the semiclassical limit(@xmath111 limit ) .
thus the zone center mode will also become gapless in the semiclassical limit and our theory can reproduce the result of the lswt in this limit .
this analysis also indicate that the gap of the zone center mode is due to quantum fluctuation and may be observable for the spin @xmath112 system .
we note previous schwinger boson mean field theory on the triangular antiferromagnet(with different mean field ansatz)predict only one gapless mode@xcite .
thus our result is an improvement over these earlier ones . at finite doping ,
the gap due to quantum fluctuation is masked by the much larger gap due to the hole background . however , the gaplessness at the zone corner is still intact as long as the system remain in the coplanar state .
this provide a explicit example of protected excitation spectrum due to quantum order . here
the quantum order is nothing but the spin berry phase . in the canted state and the collinear
state(the collinear state is a special case of the canted state ) , the degeneracy between the zone corner modes is broken . from ( [ 5 ] ) , we see the splitting between the zone corner modes is @xmath113 which increase with increasing canting angle . especially , in the collinear state
in which @xmath114 , one of the zone corner mode become degenerate with the massive zone center mode . for general canting angle
there are three nondegenerate modes in which only one ( at zone corner)become gapless at zero temperature .
since @xmath115 at the gapless point , the dispersion around it is quadratic .
the quadratic dispersion is a characteristic of the ferromagnet spin wave . in our case , canting induce a weak ferromagnetic moment .
the gaplessness of the quadratic mode indicate that the long wavelength fluctuation of the weak ferromagnetic moment is unaffected by the hole background .
the excitation spectrum for a general canting angle is shown in figure 7 .
now let s summarize the results of this paper . in this paper
, we have studied the magnetic properties of the @xmath0 model on triangular lattice in light of the recently discovered superconductivity in na@xmath1coo@xmath2 system .
we formulated the problem in the schwinger boson - slave fermion scheme and proposed a sound mean field ansatz , namely the canting ansatz , for the rvb order parameters of spins . with the canting ansatz ,
we have mapped out the temperature - doping phase diagram of the model for both sign of the hopping term .
we find the prediction of the @xmath0 model differs drastically from that of lsda calculation . in our theory , there is large doping range in which the system show zero net magnetization , rather than saturated magnetization as predicted in the lsda calculation .
we show the result of lsda is unreliable in the strong coupling regime due to its neglect of electron correlation .
the most important thing found in this paper is the vital role of the spin berry phase in this geometrically frustrated system .
we find the various phases in the phases diagram are characterized(and distinguished ) by their spin berry phase , rather than any landau - like order parameter related to broken symmetry .
we find the spin berry phase is responsible for the qualitative difference in the low energy excitation spectrum of the various states in the phase diagram .
we argue that the spin berry phase in this system may provide the first explicit example for the concept of quantum order and the transition between state with different spin berry phase may serve as the first realistic example for phase transition between states with different quantum orders .
another interesting thing found in this paper is the existence of an exotic state with nonzero spin chirality but no spin ordering in a large temperature and doping range .
we find this exotic state , which break the time reversal symmetry , support nonzero staggered current loop in the bulk of the system . since this state break a discreet symmetry , its phase boundary with other phases are well defined phase transition in the landau sense .
we propose to study this exotic state in the hole doped na@xmath3tio@xmath99 system . as a by - product of this study
, we also achieve an improvement over earlier schwinger boson mean field theory on the triangular antiferromagnet .
the canting ansatz proposed in this paper lead to two branches of gapless spinon modes .
we find the small gap of the third spinon mode is caused by quantum fluctuation and our theory can reproduce the lswt result in the semiclassical limit .
finally , we find our general phase diagram is consistent with experimental result on na@xmath1coo@xmath2 system .
we find singlet pairing picture for its superconductivity may be more relevant in the small doping regime .
the authors would like to thank members of the high tc group at castu for discussion and z.y.weng for drawing our attention to the na@xmath1coo@xmath2 material .
works of t. li is supported by nsfc grant no . | we study the magnetic properties of the @xmath0 model on triangular lattice in light of the recently discovered superconductivity in na@xmath1coo@xmath2 system .
we formulate the problem in the schwinger boson - slave fermion scheme and proposed a sound mean field ansatz(canting ansatz ) for the rvborder parameters .
working with the canting ansatz , we map out the temperature - doping phase diagram of the model for both sign of the hopping term .
we find the prediction of the @xmath0 model differ drastically from that of earlier lsda calculation and there is large doping range in which the system show zero net magnetization , rather than saturated magnetization as predicted in the lsda calculation .
we show the result of lsda is unreliable in the strong coupling regime due to its neglect of electron correlation .
we find the spin berry phase play a vital role in this geometrically frustrated system and the various states in the phases diagram are characterized(and distinguished ) by their respective spin berry phase , rather than any landau - like order parameter related to broken symmetry .
we find the spin berry phase is responsible for the qualitative difference in the low energy excitation spectrum of the various states of the phase diagram .
we argue the phase boundary in the mean field phase diagram may serve as the first explicit and realistic example for phase transition between states with different quantum orders which in our case is nothing but the spin berry phase .
we also find an exotic state with nonzero spin chirality but no spin ordering is stable in a large temperature and doping range and find this state support a nonzero staggered current loop in the bulk of the system . we propose to study this exotic state in the hole doped na@xmath3tio@xmath2 system . as a by - product of this study
, we also achieve an improvement over earlier schwinger boson mean field theory of the triangular antiferromagnet by obtaining two gapless spinon modes .
we find the small gap of the third spinon mode is caused by quantum fluctuation and our theory reproduce the linear spin wave theory(lswt ) result in the semiclassical limit . |
the noise induced relaxation in quantum systems is very important process , which is investigated in various brunches of physics and chemistry : magnetic resonance @xcite , quantum optics , nonlinear spectroscopy @xcite , solid state physics @xcite , etc .
some of these processes are analyzed assuming fast decay of correlation functions of this noise @xmath10 and considering the short correlation time limit ( sctl ) .
this approach is well known in the general relaxation theory @xcite . in the absence of memory the relaxation
is described by very popular bloch - type equations .
nevertheless , in a large number of other processes the memory effects come to play .
important steps beyond the conventional short correlation time approximation can be made within the approach treating the relaxation kinetics with the stochastic liouville equation ( sle ) @xcite .
this approach , however , also has strong limitations : it assumes the noise fluctuations to be markovian .
the alternative approach , which can be applied for the analysis memory effects is based on the zwanzig projection operator formalism @xcite .
unfortunately , in reality the projection formalism allows for tractable analysis of the memory effects only in the lowest orders in the fluctuating interaction @xmath11 inducing relaxation @xcite . at present especially significant interest to the correct interpretation of long memory ( non - markovian )
effects has been excited by recent works concerning processes governed by noises whose correlation functions @xmath12 decay anomalously slowly : @xmath13 with @xmath2 .
a large number of different phenomena , in which these processes play important role , are considered in the review article @xcite .
some of them are discussed in relation to spectroscopic studies of quantum dots shim , silb .
similar problems are analyzed in the theory of stochastic resonances @xcite .
many works study the manifestation of non - markovian long memory effects in dielectric relaxation and linear dielectric response @xcite ( and references therein ) .
all anomalous relaxation processes mentioned above can not be properly described by any methods based on the short correlation time approximation .
the conventional sle approach is not appropriate for description of these processes either .
the efficient method of analyzing the memory effects ( including the case of anomalously long memory ) has recently been developed in ref .
it treats the noise within the continuous time random walk approach ( ctrwa ) @xcite , describing the statistical properties of the noise in terms of the probability distribution functions ( pdfs ) of renewals of the interaction fluctuations associated with the noise .
the developed ctrwa - based method is based on the markovian representation of the any ctrw - processes by the markovian ones proposed in ref .
this representation allows for obtaining the formal matrix expression for the evolution operator of the system under study , which can be considered as the non - markovian generalization of the conventional sle @xcite and is called hereafter the non - markovian sle .
the non - markovian sle appeared to be very fruitful for description of some classical processes assisted by stochastic anomalous spatial migration shu2 .
the memory effects are known to manifest themselves in this processes extremely strongly leading to strongly non - exponential anomalous relaxation kinetics @xcite . in this work
the proposed non - markovian sle is applied to the detailed analysis of specific features of relaxation in quantum systems induced by anomalous noise , i.e. the noise with anomalously long - tailed correlations in time , whose effect on quantum systems can hardly be correctly treated by expansion in powers of the noise amplitude .
this type of the sle enables one to describe relaxation kinetics without expansion in the fluctuating interaction . in some physically reasonable approaches , for example in the sudden relaxation model ( srm )
, it allows for describing the phase and population relaxation kinetics in analytical form . in the work
two types of processes are considered : relaxation in two - level systems ( tls ) and anomalous - diffusion assisted radical pair ( rp ) recombination in a potential well , as an example of processes in multilevel systems .
relaxation in these systems is shown to be strongly non - exponential .
in addition , in rp recombination anomalous properties of relaxation ( caused by anomalous relative diffusion ) result in some peculiarities of magnetic field effects @xcite : non - analytical dependence of magnetic field affected recombination yield ( mary ) on the parameters of the spin hamiltonian , strongly non - lorenzian shape of lines of reaction yield detected magnetic resonance ( rydmr ) @xcite , etc .
we consider noise induced relaxation in the quantum system whose evolution is governed by the hamiltonian @xmath14 where @xmath15 is the term independent of time and @xmath16 is the fluctuating interaction , which models the noise .
the evolution is described by the density matrix @xmath17 satisfying the liouville equation ( @xmath18 ) @xmath19 = [ h \rho - \rho h ] .
\label{gen2}\ ] ] @xmath20-fluctuations are assumed to be symmetric ( @xmath21 ) and result from stochastic jumps between the states @xmath22 in the ( discrete or continuum ) space @xmath23 with different @xmath24 and @xmath25 ( i.e. different @xmath26 $ ] and @xmath27 ) : @xmath28 hereafter we will apply the bra - ket notation : @xmath29 for the eigenstates of @xmath30 ( in the original space ) and @xmath31 ( in the liouville space ) , and for states in @xmath32-space , respectively .
it is worth noting that within the proposed semiclassical approach one can also describe some fluctuating irreversible processes in the system modelled by the additional non - hermitian relaxation term @xmath33 in the lioville operator @xmath34 . in this model
the effect of fluctuating irreversible process results in modification of the liouville equation ( gen2 ) : @xmath35 in the majority of processes the time evolution of observables under study is determined by the evolution operator @xmath36 in the liouville space averaged over @xmath16- and @xmath37-fluctuations : @xmath38 the operator @xmath36 can equivalently be represented in terms of the conditional evolution operator @xmath39 averaged over the initial distribution @xmath40 : @xmath41 for steady state @xmath42-fluctuations the averaging should be made over the equilibrium distribution @xmath43 , i.e. @xmath44 .
it is worth noting , however , that evaluation of some observables requires analysis of the conditional operator @xmath45 itself rather than the averaged one @xmath36 ( see sec .
thus , in the proposed semiclassical approximation for @xmath42- and @xmath37-fluctuations the relaxation kinetics is determined by the operator @xmath45 .
its evaluation is , in general , a complicated problem . in what follows
we will discuss important approaches in which this problem can be significantly simplified by reducing it to solving the ( differential or integral ) sle for @xmath45 @xcite . in our further study we will often use the laplace transformation of functions under study in time @xmath46 conventionally denoted as @xmath47 for any function @xmath48 .
the goal of our work is to analyze the strong memory effects within the ctrwa .
this approach is known to result in the complicated non - markovian sle @xcite for @xmath45 .
however , it is instructive to start the discussion with the simpler markovian models reducing to the conventional semiclassical sle . in the markovian approach the kinetics of jumps in @xmath32-space leading to @xmath42-fluctuations
are described by the pdf @xmath49 satisfying equation @xcite @xmath50 where @xmath51 is some linear operator . the principal simplification of the problem results from the fact that in the markovian approach ( [ mark1 ] ) the @xmath52 obeys the sle @xcite : @xmath53 so that in accordance with eq . ( [ gen5b ] )
we get for the laplace transform @xmath54 : @xmath55 the idea of the sle ( [ mark2 ] ) is based on the simple observation that the changes of @xmath56 resulted from dynamical motion [ eq .
( [ gen2 ] ) ] and stochastic evolution [ eq .
( [ mark1 ] ) ] during short time @xmath57 are written as @xmath58 respectively , so that total change of @xmath59 is given by the expression @xmath60 equivalent to eq .
( [ mark2 ] ) .
non - markovian @xmath16-fluctuations can conveniently be described by the ctrwa [ which leads to the non - markovian sle @xcite for @xmath61 .
it treats fluctuations as a sequence of sudden changes of @xmath62 .
the onset of any particular change of number @xmath63 is described by the matrix @xmath64 ( in @xmath65-space ) of probabilities not to have any change during time @xmath46 and its derivative @xmath66 , i.e. the pdf - matrix for times of waiting for the change .
these matrices are diagonal and independent of @xmath63 at @xmath67 : @xmath68 for @xmath69 @xmath70 depend on the problem considered . for non - stationary ( @xmath71 ) and
stationary ( @xmath72 ) fluctuations @xcite @xmath73 respectively , where @xmath74 is the matrix of average times of waiting for the change scher1,scher2,wei .
it is worth noting some relations for the laplace transforms of @xmath75 and @xmath76 suitable for our further analysis : @xmath77/\epsilon$ ] and @xmath78 @xcite , as well as the representations @xmath79^{-1 } \;\ : \mbox{and } \;\ : \hat { \widetilde p } ( \epsilon ) = [ \epsilon + \epsilon/\hat \phi ( \epsilon)]^{-1}. \label{nmark4}\ ] ] in terms of the auxiliary matrix @xmath80 diagonal in @xmath81-space ( see below ) . in this section
we will briefly discuss the markovian representation of the ctrwa recently proposed in ref .
@xcite , which appears to be very useful for the ctrwa - based analysis of the problem under study and , in particular , provides the most rigorous method of deriving the non - markovian sle .
suppose that the kinetics of @xmath82-transitions in @xmath83-space is controlled by the markovian process in another @xmath84-space , which is governed by the operator @xmath85 .
the corresponding pdf @xmath86 satisfies equation @xmath87 describing evolution in @xmath88-space and equilibration if the operator @xmath85 has the equilibrium state @xmath89 ( @xmath90 where @xmath91 are the probabilities of equilibrium population of states in @xmath84-space .
the @xmath92-process is assumed to control the evolution in @xmath83-space as follows : @xmath82-transitions occur with the rate @xmath93 whenever the system visits the transition state @xmath94 in @xmath88-space .
transitions can lead to the change in @xmath95-state , i.e to @xmath96transition with @xmath97 . for simplicity , we assume that @xmath98 as well as that @xmath85 and @xmath99-state are independent of the state in @xmath65-space .
the evolution of the system in @xmath100space is described by the pdf matrix @xmath101 obeying the sle @xmath102 where @xmath103 are the transition matrices ( operating in @xmath104-space ) are diagonal and non - diagonal in @xmath65-subspace , respectively , in which @xmath105 and @xmath106 .
equation ( [ ma2 ] ) should be solved with the initial condition @xmath107 where @xmath108 .
the function of interest is the ( time ) laplace transformed pdf in @xmath32-space @xmath109 it is determined by solution @xmath110 of equation @xmath111 where @xmath112 the expression for @xmath113 , obtained by solving eq .
( [ ma7 ] ) , can be represented in a ctrwa - like form , which in what follows [ by analogy with the markovian model ( [ mark1])-([mark2a ] ) ] will be treated as the solution of the so called non - markovian sle shu1 .
hereafter , for brevity , we will sometimes omit the argument @xmath114 of the laplace transforms of functions under study if this does not result in confusions .
solution of eq .
( [ ma7 ] ) leads to the following non - markovian sle , written in matrix resolvent form @xcite , @xmath115^{-1 } \hat { \mathcal{p}}\hat { \widetilde w}_{i } ( \hat \omega ) , \label{ma9}\ ] ] where @xmath116 is the matrix of jump probabilities with zeroth diagonal elements and @xmath117 are written as @xmath118 in which @xmath119 the non - markovian sle ( [ ma9 ] ) can also be represented in an alternative convenient form @xcite @xmath120 where @xmath121 is the kolmogorov - feller - type operator describing the jump - like motion in @xmath65-space , @xmath122 is the matrix function ( diagonal in @xmath65-space ) characterizing @xmath123 [ see eq . ( [ nmark4 ] ) ] : @xmath124 , and @xmath125 are auxiliary ( modified ) pdf matrices of the type of waiting time pdfs .
the representation ( [ ma17 ] ) of the non - markovian sle is somewhat different from that proposed in ref .
both representation are , however , equivalent and can equally be used for the analysis , though eq .
( [ ma17 ] ) is closer in its form to the ctrwa - like expressions @xcite .
the operator @xmath51 is assumed to have the equilibrium eigenstate @xmath126 ( i.e. @xmath127 ) : @xmath128 the eigenstate @xmath126 , however , is not the true equilibrium state of the system under study .
the true equilibrium state @xmath129 is also determined by the behavior of @xmath80 at @xmath130 .
it is clear from eq .
( [ ma17b ] ) and definition of the pdf matrix @xmath131 that @xmath132 , therefore , in general , we can write @xmath133 in this equation the matrices @xmath134 ( of exponents ) and @xmath135 [ of characteristic rates ( see sec
. v ) ] are assumed to be diagonal in @xmath32-space .
moreover , for simplicity , to avoid analysis of exotic equilibrium states @xcite we assume that @xmath136 is just a parameter independent of @xmath137 rather than matrix . in accordance with formulas ( [ ma17])-([ma17b ] ) , for such @xmath138 @xmath139 ,
i.e. the state @xmath129 is , actually , the equilibrium eigenstate of @xmath140 ( @xmath141 ) , which is written as @xmath142 possible expressions for @xmath129 in some particular models of @xmath51 are discussed below in sec .
it is important to note that with the eigenstate @xmath129 the average of any operator @xmath143 can be represented as @xmath144 in particular , as it follows from eq .
( [ gen5a ] ) , @xmath145 according to eqs .
( [ ma9])-([ma16a ] ) the initial state @xmath146 ( in @xmath147 ) manifests itself only in the expressions for matrices @xmath148 and @xmath149 . in particular @xcite : \a ) in the non - stationary @xmath71-ctrwa @xmath150 , so that @xmath151 , @xmath152 and @xmath153^{-1}. \label{ma18}\ ] ] \b ) in the stationary @xmath154-ctrwa @xmath155 , where @xmath156 is the equilibrium eigenstate [ see eq .
( [ ma1a ] ) ] , consequently @xmath157 where @xmath158 is the matrix of average times ( diagonal in @xmath32-space ) with @xmath159 . in the considered case @xmath160 ,
when @xmath161 and @xmath162 .
substitution of @xmath163 into eq .
( [ ma9 ] ) yields @xmath164
the srm @xcite assumes sudden equilibration in @xmath65-space described by operator @xmath165 in which @xmath166 is some auxiliary vector determined by the equilibrium vector @xmath129 : @xmath167 where @xmath168$}. \label{sud2a}\ ] ] according to eqs .
( [ sud2 ] ) and ( [ sud2a ] ) : @xmath169 this relation determines the vector @xmath170 ( in the definition of @xmath51 ) which ensures the given equilibrium state @xmath129 . the value of the parameter @xmath171 is fixed by the normalization condition for the distribution @xmath172 : @xmath173 = 1 .
\label{sud2c}\ ] ] in the model ( [ sud1 ] ) one gets for any @xmath174 @xmath175^{-1 } \langle \hat { \widetilde{w}}_{\!q_i } \rangle , \label{sud3}\ ] ] where @xmath176 , @xmath177 the obtained general formulas are simplified in some particular models , for example , in the @xmath178-state srm with @xmath179 in this model @xmath180 .
it is important to note that in the case of two states the model ( [ sud5 ] ) is the only possible one .
particularly simple expressions are obtained in the model of the only characteristic fluctuation rate : @xmath181 for @xmath154-fluctuations this model assumes the average time @xmath182 to be independent of @xmath137 : @xmath183 .
together with the model ( [ sud5 ] ) it implies equipopulated equilibration with @xmath184 so that @xmath185 it predicts , for example , for @xmath71-fluctuations ( @xmath186 ) @xmath187^{-1}. \label{sud7}\ ] ] the model ( [ sud5a ] ) clearly reveals all important specific features of the non - markovian - noise - induced relaxation in the simplest form .
another simple model allowing for analytical consideration of problem is the diffusion model . in this work
the diffusion model will be applied to the analysis of recombination of the pair of radicals assuming that one of radicals undergoes isotropic diffusion in three dimensional space , say @xmath188-space , while another radical does not move and is located at @xmath189 .
conventionally , the diffusive motion of the moving radical is described by the smoluchowski - type jump operator @xmath190 @xcite . for simplicity
we will consider isotropic processes for which @xmath191 , \label{dif1}\ ] ] where @xmath192 , @xmath193 is the gradient operator , @xmath194 is the average square of the jump length independent of @xmath195 , and @xmath196 is the external interaction potential .
the potential @xmath197 is assumed to be of the shape of ( deep ) spherically symmetric square well with the radius @xmath198 much larger than the distance @xmath199 of closest approach : @xmath200 with @xmath201 .
it worth noting that within the continuum model of stochastic jumps resulting in @xmath42- and @xmath37-fluctuations the corresponding operators @xmath202 and @xmath203 are just functions of @xmath204 . in the considered spherically symmetric
case @xmath205 in practical applications of special importance is the sctl for @xmath16-fluctuations in which eq .
( [ sud3 ] ) can markedly be simplified .
it corresponds to the large characteristic ( correlation ) rate @xmath206 of fluctuations , i.e. the large characteristic time of the dependence @xmath207 : @xmath208 . in our analysis
we will discuss the general class of models for which @xmath209 is represented in the form @xmath210 .
for such models the correlation rate can naturally be defined as @xmath211 . in the sctl the relaxation kinetics
is described by the first terms of the expansion of @xmath212 in small @xmath213 , since @xmath80 is the increasing function of @xmath214 with @xmath215 [ see sec .
iib and , in particular , eq . ( [ ma17d ] ) ] .
this fact allows one to significantly simplify the problem under study .
nevertheless , some important general conclusions can be made independently of the form of @xmath216 as it will be shown in sec .
v. here we describe two systems which will be analyzed to illustrate the obtained general results : ( 1 ) the isolated quantum two - level system ( tls ) , whose relaxation results from @xmath42-fluctuations described within the stochastic two - state model , and 2 ) the geminate pair of radicals diffusing in the potential well in which spin evolution , affected by the fluctuating reactivity @xmath217 , in turn strongly influences the reactivity .
[ [ two - level - model . ] ] two - level model .
+ + + + + + + + + + + + + + + + quantum evolution of the tls is governed by the hamiltonian ( assumed to be real matrix ) with @xmath218 where @xmath219 \;\;\mbox{and } \;\ ; \sigma_x = \left[\begin{array}{cc } 0 & 1 \\ 1 & 0 \\ & \end{array}\right ] \ : \begin{array}{c } \end{array}. \label{two1}\ ] ] [ [ two - state - model - of - fluctuations . ] ] two - state model of fluctuations .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the two - state model of fluctuations @xmath16-modulation is assumed to result from jumps between two states ( in @xmath220-space ) , say , @xmath221 and @xmath222 .
it is important to note that in the particular case of two states any ctrwa - based kinetics model reduces to the simple two - state srm [ see eqs .
( [ sud5 ] ) and ( [ sud5a ] ) ] in which @xmath223 , and @xmath224 [ [ simple - variants - of - tls - and - two - state - model . ] ] simple variants of tls and two - state model .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + below we will consider two examples of these models : \1 ) @xcite : @xmath225 , @xmath226 and @xmath227 \2 ) : @xmath228 and @xmath229 , so that @xmath230 in our further analysis the first model is applied to the description of dephasing while the second one is used in studying population relaxation .
[ [ calculated - observables . ] ] calculated observables .
+ + + + + + + + + + + + + + + + + + + + + + + in the model ( [ two1 ] ) dephasing and population relaxation can be characterized by two functions : \1 ) the spectrum @xmath231 which is taken in the form corresponding to fourier transformed free - induction - decay ( ftfid ) experiments @xcite @xmath232 \2 ) the difference of level populations @xmath233 in these two functions @xmath234 [ [ magnetic - field - effects . ] ] magnetic field effects .
+ + + + + + + + + + + + + + + + + + + + + + + the kinetics of rp recombination is known to be markedly affected by the rp spin evolution which is controlled by the spin hamiltonian @xmath30 of the pair . the dependence of the recombination kinetics on rp spin state results in a large number of phenomena called magnetic field effects @xcite . in this work
we will restrict ourselves to some simple and representative effects observed in strong magnetic field @xmath235 for which the zeeman interaction is much larger than the intraradical magnetic interactions ( hyperfine interaction , etc . ) .
we will also consider the effect of the external microwave field @xmath236 rotating with the frequency @xmath8 in the plane perpendicular to the vector @xmath235 .
for strong magnetic fields and in the presence of the field @xmath236 the spin hamiltonian governing spin evolution of electrons in the pair of radicals , say @xmath237 and @xmath238 , can conveniently be written in the frame of reference rotating together with @xmath236 with the frequency @xmath8 @xcite : @xmath239 where @xmath240 and @xmath241 is the zeeman frequency of the radical @xmath242 possessing some paramagnetic nuclei with hyperfine interactions @xmath243 and @xmath244 . in case of need the matrix representation of the hamiltonian ( [ rp1 ] )
can be determined either in the bases of radical spin states @xmath245 or in the basis of eigenstates of the total electron spin @xmath246 : singlet ( @xmath247 ) and triplet ( @xmath248 ) states , which are expressed as @xmath249 ; the rp recombination can be treated as a contact reaction at a distance of closest approach @xmath199 using the simple model @xcite @xmath250 where @xmath251 is anticommutator ( @xmath252 ) in which @xmath253 is the operator of projection on the singlet ( @xmath247 ) spin state of rp .
the rp is assumed to be initially created in the singlet state @xmath254 within the potential well at a distance @xmath255 ( @xmath256 ) , so that @xmath257 [ [ observables . ] ] observables .
+ + + + + + + + + + + + in experiments on magnetic field effects a number of observables are discussed @xcite . here
we analyze the most simple ones : \1 ) magnetically affected reaction yield ( mary ) @xcite , measured in the external constant magnetic field ,
\2 ) reaction yield detected magnetic resonance ( rydmr ) @xcite , i.e. microwave field influenced recombination yield . in both types of experiments
the observables under study are recombination ( @xmath7 ) and dissociation ( @xmath258 ) yields : @xmath259 \label{rp4}\ ] ] and @xmath260
. naturally , the expression ( [ rp4 ] ) should be averaged over nuclear configurations ( over @xmath261 ) . however , in our further discussion we will omit this evident procedure and analyze the behavior of @xmath7 for fixed @xmath262 and @xmath263 ( note that the case of fixed @xmath262 and @xmath263 can be realized experimentally , for example , with rps which do not contain paramagnetic nuclei ) .
[ [ small - h_s - w_c - ll-1 . ] ] small @xmath264 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + within the sctl relatively simple and general results can be obtained in the case @xmath264 . in the lowest order in @xmath265 @xcite @xmath266 this formula holds for any initial matrix @xmath174 and , in particular , for @xmath154-fluctuations , if @xmath267 .
it can easily be obtained from eq .
( [ sud3 ] ) if one takes into account that @xmath268 .
[ [ large - h_s - w_c - gtrsim-1 . ] ] large @xmath269 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + somewhat more complicated sctl - case @xmath270 can be analyzed by expanding @xmath113 in powers of the parameter @xmath271 .
in particular , within the general two - level model [ eq .
( [ two1 ] ) ] with @xmath272 in the second order in @xmath273 the diagonal and non - diagonal elements of @xmath17 are decoupled and the corresponding elements of @xmath274 are expressed in terms of the universal function @xmath275 where @xmath276 and @xmath277 \mathcal{v}_n \rangle . \label{sctl2}\ ] ] the simplest model for anomalous fluctuations can be written as @xcite @xmath278 where @xmath135 is the matrix of characteristic fluctuation ( correlation ) rates diagonal in @xmath279-basis .
for the sake of simplicity , @xmath135 is assumed to be independent of @xmath137 , i.e. @xmath280 so that one can use formula ( [ sud7 ] ) for evaluation of @xmath281 .
the model ( [ res1 ] ) describes anomalously slow decay of the pdf - matrix @xmath282 ( very long memory effects in the system @xcite ) , for which only the case of non - stationary ( @xmath71 ) fluctuations is physically sensible . for small @xmath264
( [ sctl1 ] ) ] the model ( [ res1 ] ) yields the expression constituting the important result of the work : @xmath283 this expression demonstrates the surprising property of relaxation induced by anomalous noise in the sctl : the evolution operator @xmath284 [ and @xmath285 , which determines relaxation kinetics , is independent of the characteristic rate @xmath3 .
it is also worth noting that for @xmath286 and @xmath287 formula ( res2 ) describes relaxation kinetics corresponding to the static and fluctuation narrowing limits @xcite in which @xmath288 respectively . at intermediate values @xmath289 the kinetics
is represented by a non - trivial combination of the static- and narrowing - like expressions whose relative contribution is determined by @xmath6 , as it is seen from eq .
( [ res2 ] ) . of certain interest is the limit @xmath290 in which formula ( [ res2 ] ) predicts the bloch - type exponential relaxation : @xmath291^{-1}. \label{res2a}\ ] ] the relaxation is controlled by the rate matrix @xmath292 , and is accompanied by frequency shifts represented by @xmath293 .
the peculiarity of anomalous relaxation , however , shows itself in the independence of matrices @xmath294 and @xmath295 ( unlike those in the conventional bloch equation ) from the characteristic rate @xmath3 of @xmath42-fluctuations . in the simple model ( [ two3 ] ) of diagonal noise the spectrum @xmath296 can be obtained in analytical form in relatively general assumptions on @xmath297-fluctuations .
[ [ general - srm - and - sctl . ] ] general srm and sctl .
+ + + + + + + + + + + + + + + + + + + + + in the sctl ( i.e. for large rate @xmath3 ) within the general srm ( [ sud1 ] ) relatively simple expression for @xmath296 can be derived without any assumption on the structure of energy levels ( for any number of states ) shu4 : @xmath298 where @xmath299\rangle \label{res3a}\ ] ] with @xmath300 being the heaviside step - function . in general , this formula is too cumbersome for studying the specific features of the spectrum . much more clearly they can be revealed with the use of the two - state srm ( see below ) .
[ [ arbitrary - rate - w - in - two - state - srm . ] ] arbitrary rate @xmath3 in two - state srm .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the two - state srm ( [ two2 ] ) allows for the analytical analysis of the spectrum @xmath296 for any value of @xmath3 , i.e. outside the region of applicability of the sctl : @xmath301 in this expression @xmath302 where @xmath303.\;\;\ ; \label{res4e}\end{aligned}\ ] ] figures 1 and 2 demonstrate a large variety of dependencies @xmath296 for different values of the parameter @xmath304 . in general , the spectrum possesses two peaks at @xmath305 which at @xmath306 and @xmath290 collapse into one central peak ( see next paragraph ) .
since in accordance with the definition ( sec .
ivb ) the limit @xmath306 corresponds to the sctl , the collapse of lines for @xmath290 looks quite natural @xcite . simple analysis of eq .
( [ res4c ] ) and calculation results displayed in fig .
1 show that in this limit general spectrum ( [ res4c ] ) reduces to the limiting one confined in the region @xmath307 [ see eq .
( [ res4 ] ) and its discussion ] . with the increase of @xmath273 ,
however , the delocalization of the spectrum outside this region is predicted ( fig .
2 ) . the delocalization becomes especially pronounced in the limit @xmath290 , as expected .
[ [ sctl - within - two - state - srm . ] ] sctl within two - state srm .
+ + + + + + + + + + + + + + + + + + + + + + + + + + in the two - state srm ( [ two2 ] ) formula ( [ res3 ] ) is significantly simplified to give the same result as eq .
( [ res4c ] ) in the limit @xmath308 : @xmath309where @xmath310 ( see also ref .
silb ) . according to this formula anomalous dephasing ( unlike conventional one @xcite ) leads to broadening of @xmath311 only in the region @xmath312 and singular behavior of @xmath311 at @xmath313 : @xmath314 . for @xmath315
[ @xmath316 satisfies the relation @xmath317 the formula also predicts the occurrence of the central peak ( at @xmath318 ) @xcite of lorenzian shape and width @xmath319@xmath320 , \label{rs4b}\]]whose intensity increases with the increase of @xmath321 ( see fig .
1a ) . at @xmath322
the parameters of this peak are reproduced by eq .
( [ res2a ] ) in which @xmath323 .
the origination of the peak indicates the transition from the static broadening at @xmath324 to the narrowing one at @xmath325 [ see eq .
( [ res2 ] ) ] .
it is worth noting that , of course , for systems with complex spectra this transition can strongly be smoothed and almost indistinguishable experimentally .
[ [ dephasing - in - sctl - for - hw - ll-1 . ] ] dephasing in sctl for @xmath326 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the two - state srm ( [ two2 ] ) enables one to analyze the behavior of the spectrum @xmath231 in the complicated case @xmath327 ( within the applicability region for the sctl ) . in this case
dephasing is determined by eq .
( [ res2 ] ) .
after some manipulations one arrives at @xmath328 , \label{deph1}\ ] ] where @xmath329 , \label{deph2 } \\ & & r_0 = 2v/\omega_s , \;\ ; \mbox{and } \;\
; e_0 = ( \omega_s/2 ) \sqrt{1+r_0
^ 2}. \label{deph2a}\end{aligned}\ ] ] the spectrum @xmath330 predicted by eq .
( [ deph1 ] ) is depicted in fig .
3 as a function of @xmath331 for two values of @xmath6 and different values of the parameter @xmath332 . this figure demonstrates non - trivial specific features of the shape of @xmath231 depending on the values of @xmath6 and @xmath332 .
first of all , similarly to the case of diagonal noise , ( in the limit @xmath327 ) the spectrum @xmath231 is localized in the region @xmath333 for all values of @xmath6 . outside this region @xmath334 .
the behavior of @xmath231 significantly changes with the increase of @xmath6 : \1 ) at small @xmath335 the spectrum consists of three peaks ( see fig .
3a ) : the central peak at @xmath336 and symmetric edge peaks at @xmath337 ( at the spectrum edges ) , in vicinity of which the behavior of @xmath338 , where @xmath339 .
the intensity of the central peak decreases as @xmath6 is increased .
\2 ) for large @xmath340 each of two symmetric peaks at @xmath341 split into two ones , so that the spectrum possesses five peaks : at @xmath336 , at @xmath342 and at @xmath343 with @xmath344 ( see fig .
moreover , the intensity of three original edge and central peaks ( at @xmath342 and @xmath345 ) decreases as @xmath290 while the intensity two additional peaks increases . besides , the two peaks at @xmath343 approach each other , i.e. @xmath346 with the increase of @xmath332 ( as it is shown in fig .
3b ) , and in the limit @xmath347 collapse into one peak . [
[ dephasing - in - sctl - for - large - protectomega_sw - gtrsim-1 . ] ] dephasing in sctl for large @xmath348 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the model ( [ two4 ] ) reveals some additional specific features of the kinetics of phase relaxation in the case of not very large @xmath3 , when @xmath349 .
for example , as it is seen from eqs .
( [ sctl4a ] ) and ( [ sctl4b ] ) , in the limit @xmath350 matrix elements @xmath351 which describe phase relaxation are given by @xmath352 , \label{res50}\ ] ] where @xmath353 is the mittag - leffler function @xcite . in this case the spectrum @xmath354 and @xmath355 with @xmath356 , @xmath357 and @xmath358 .
\label{res5c}\ ] ] formulas ( [ res5])-([res5c ] ) predict singular behavior of @xmath231 at @xmath359 : @xmath360 , and slow decrease of @xmath231 with the increase of @xmath361 : @xmath362 . in the limit @xmath363 @xmath364 so that @xmath365 .
this means that for @xmath366 the spectrum @xmath231 is localized in the region @xmath367 and looks similar to @xmath231 for diagonal dephasing at @xmath368 and @xmath306 ( see fig .
for @xmath369 , however , @xmath231 is non - zero outside this region as well , moreover , in the limit @xmath370 the spectrum @xmath371 becomes symmetric : @xmath372 .
such dependence of @xmath231 on @xmath373 is also very similar to @xmath231-dependence on @xmath273 found above in the case of diagonal noise .
it is interesting to note that for @xmath374 functions @xmath375 and @xmath296 are independent of @xmath3 [ in agreement with eq .
( [ res2 ] ) ] since @xmath376 and @xmath377 . in the opposite limit , however , @xmath378 so that the characteristic relaxation time @xmath379 .
specific features of anomalous population relaxation can be analyzed with the model of non - diagonal noise ( [ two4 ] ) .
in particular , in the limits @xmath350 and @xmath380 with the use of eqs .
( [ sctl4a]),([sctl4b ] ) and ( [ res2a ] ) one gets @xmath381 \;\;\mbox{and } \;\ ; n(t ) = e^{- w_{\alpha } t } , \label{res6}\ ] ] respectively , where @xmath382 is the mittag - leffer function defined above and @xmath383 .
the first of these formulas predicts very slow population relaxation at @xmath384 : @xmath385 .
similar to @xmath296 the function @xmath386 is , in fact , independent of @xmath3 for @xmath387 because in this limit @xmath388 . in the opposite limit
@xmath389 the characteristic time population relaxation is @xmath379 since @xmath390 is independent of @xmath3 as in the case of phase relaxation . for @xmath391
one obtains @xcite @xmath392 where @xmath393 and @xmath394 are defined in eqs .
( [ deph2 ] ) and ( [ deph2a ] ) , respectively .
it is easily seen that in the corresponding limits the expression ( [ res7 ] ) reproduces formulas ( [ res6 ] ) with @xmath395 and @xmath396 . outside these limits
@xmath386 can be evaluated numerically ( some results are shown in fig .
4 ) . in general
, @xmath386 is the oscillating function ( of frequency @xmath397 ) with slowly decreasing average value and oscillation amplitude : for @xmath398 @xmath399 ( except the limit @xmath290 ) . at large @xmath400 one can estimate the asymptotic behavior of @xmath386 : @xmath401 \frac{1}{\tau^{\alpha}}.\quad \label{res7a}\end{aligned}\ ] ] the kinetic dependencies found in this section for @xmath402 and expressed in terms of the mittag - leffler function are conveniently represented in the form of the equation similar to the conventional bloch equation for the density matrix but with the fractional derivatives .
it is easily seen that the kinetic functions ( [ res50 ] ) and ( [ res6 ] ) can be considered as solution of the equation @xmath403_{s}^{\!\times } \hat { \mathcal{r } } , \label{res8}\ ] ] where @xmath404_{s}^{\!\times}\!f = \frac{1}{\gamma ( \alpha ) } e^{-i\hat h_s t}\frac{\partial}{\partial t } \ !
\int_0^t \!\ ! d\tau \frac{e^{i\hat h_s \tau}}{(t\!-\!\tau)^{1\!-\alpha } } f ( \tau ) \label{res9}\ ] ] is the modified liouville - riemann fractional integral operator , and @xmath405 this equation generalizes the well known classical expressions @xcite to the quantum processes .
the analysis presented above shows that the effect of the anomalous noise , i.e. fluctuating interaction [ whose correlation functions is anomalously long tailed : @xmath13 with @xmath289 ] , on quantum systems can be very strong , manifesting itself in anomalous relaxation kinetics . with the increase of @xmath6 up to @xmath407
the effects of anomaly of interaction fluctuations become weaker but , nevertheless , they still manifest themselves in the relaxation kinetics . to clarify these effects
we will briefly discuss the model in which @xmath408 where @xmath289 , and @xmath3 and @xmath409 are the constants with @xmath410 [ small value of @xmath409 ensures that @xmath411 .
this @xmath412 corresponds to the waiting time pdf - matrix @xmath131 for which the average time @xmath413 is finite but the higher moments @xmath414 with @xmath415 do not exist .
possible effects of this weakly anomalous noise can be analyzed within the sctl with the use of eqs .
( [ sctl1])-([sctl4b ] ) .
for example , in the limit @xmath326 one obtains formula @xmath416^{-1}. \label{res10}\ ] ] which predicts the bloch - type relaxation of both phase and population , but with the rate @xmath417 depending on the @xmath3 as @xmath418 , i.e. slower than in the conventional bloch equation ( @xmath419 @xcite ) .
more detailed analysis also demonstrates that in this expression for @xmath420 the terms @xmath421 are also expected .
they lead to the inverse power - type asymptotic behavior of @xmath422 which , however , is observed only at very long times @xmath423 .
this conclusion can easily be clarified with the use of the general formula ( [ sctl4a ] ) valid in the case @xmath424 for any model of @xmath412 . according to this formula for the model ( [ res10a ] ) the kinetics of both phase and population relaxation
is similar in its mathematical form to the probability of fluctuations @xmath12 [ see eq .
( nmark1 ) ] which is , evidently , of power - type behavior at long times : @xmath425 , in agreement with the above statement .
formula ( [ sctl4a ] ) allows for making the following general conclusion on the relaxation kinetics : the kinetics is of anomalous long tailed inverse - power - type behavior for any ctrwa - based model of fluctuations assuming singular @xmath426 with brunching points . in accordance with eq .
( [ rp4 ] ) , the observables investigated in mary and rydmr experiments are expressed in terms of the operator @xmath427 . within the ctrwa equation for this operator
is determined by the model of migration of the mobile radical . here
we will consider the anomalous diffusion model of migration in which the memory is anomalously long and is described by the operator @xmath428 ( [ res1 ] ) .
notice that in the considered process of spatial diffusion the ( diagonal ) matrix of characteristic rates @xmath135 in the operator @xmath80 is actually represented as a distance dependent function : @xmath429 . for simplicity , in our further discussion of radical diffusion
we assume the rate @xmath430 and the exponent @xmath431 to be independent of @xmath332 : @xmath432 and @xmath433 . in this case
the ctrwa , with @xmath428 given by ( [ res1 ] ) , is known to predict anomalous diffusion @xcite . the considered anomalous diffusion of the mobile radical of the rp , evidently , results in non - stationary fluctuations of reactivity @xmath434 .
the effect of these fluctuation is described by the operator @xmath427 satisfying the non - markovian sle ( [ ma18 ] ) which can be represented in terms of the so called fractional diffusion equation ( for the laplace transform ) as follows @xmath435 where @xmath436 and @xmath437 in general , solution of the non - markovian sle ( [ mfe1 ] ) with the smoluchowski - type @xmath438 [ see eq .
( [ dif1 ] ) ] is a fairly complicated ( though , in principle , analytically tractable @xcite ) problem .
the most interesting specific features of magnetic field effects , however , can quite clearly be illustrated in some limiting cases allowing for considerable simplification of the obtained general expressions . in our analysis
we will concentrate on one of such cases , corresponding to the limit of the deep well @xmath197 and fast diffusion within the this well , which is described by the cage model @xcite .
similar to the case of conventional diffusion , fast anomalous diffusion in a large enough and deep potential well @xmath439 , for which @xmath440 and @xmath201 leads to rapid relaxation of the initial non - equilibrium spatial distribution of the radical within the well during the time @xmath441 and formation of the nearly homogeneous quasiequilibrium state ( cage ) within the well . at longer times
@xmath442 reaction and dissociation are shown to result in the quasistationary decay of this state which is , naturally , independent of the distance @xmath443 of rp creation . at these times , also as for reactions assisted by conventional diffusion , the ( anomalous ) kinetics of the process under study is described by lowest pole in the expression ( [ ma18 ] ) ( determined by the lowest eigenvalue of the operator @xmath438 ) @xcite . in what follows this approximation will be called the anomalous - cage model . keeping this lowest pole after some algebraic manipulations similar to those presented in ref .
[ 27 ] one arrives at formula @xmath444 , \label{mfe3}\ ] ] where @xcite @xmath445 and @xmath446 is the operator of projection on the singlet ( @xmath254 ) state of the rp .
formula ( [ mfe3 ] ) is quite suitable for the analysis of the problem under study .
unfortunately , in general , in the cage model ( [ mfe3 ] ) the expressions for @xmath7 are still fairly cumbersome and can be used mainly for numerical estimations . to reveal characteristic properties of @xmath7-dependence on the parameters of the model we consider simple limiting cases which will be somewhat different for mary and rydmr .
[ [ analysis - of - mary ] ] analysis of mary + + + + + + + + + + + + + + + + in the considered limit of strong magnetic field @xmath447 , in which @xmath448 , the effect of spin evolution on reaction yield called mary can be studied within the @xmath449-approximation . in this approximation , which takes into account that in the strong magnetic field limit the contribution of @xmath450-terms to the reaction yield is negligibly small , the rp spin hamiltonian is written as @xmath451 detailed analysis demonstrates that , in general , the anomalous - cage model predicts the mary - dependence on parameters of the spin hamiltonian similar to that known in the conventional cage model ( conventional diffusion assisted processes within the well ) @xcite , however , with replacement of analytical functions by non - analytical ones .
most important features of non - analytical mary - dependence on the parameters of the spin hamiltonian , predicted by the anomalous - cage model , can be demonstrated in the simple limit of relatively weak magnetic interactions : @xmath452 . in this limit
one can evaluate mary with the lowest order of expansion of @xmath7 [ eq . ( [ mfe3 ] ) ] in powers of @xmath453 : @xmath454 , \label{mfe4a}\ ] ] where @xmath455 is the reactivity in the singlet state and @xmath456 .
straightforward evaluation with the use of eq .
( [ mfe4a ] ) gives the expression @xmath457 the non - analytical dependence @xmath458 is just the manifestation of anomalous nature of diffusion within the well ( anomalous nature of the cage ) .
noteworthy is that in the case @xmath290 the @xmath459-dependent part vanishes .
this is because in eq .
( mfe4a ) the spin dependent contribution to @xmath7 is taken into account in the lowest order in @xmath460 [ @xmath461 ) ] . at @xmath287 , however , the term of this order , linear in @xmath462 , does not contribute to @xmath463 . the non - zero contribution , evidently , results only from the second order term . [ [ analysis - of - rydmr ] ] analysis of rydmr + + + + + + + + + + + + + + + + + consideration of the most important specific features of rydmr can significantly be simplified in the limit of large @xmath449-coupling @xmath464 [ see eq .
( rp1 ) ] and relatively weak microwave field @xmath9 : @xmath465 and @xmath466 . in this limit quantum coherence effects on evolution of all states with large
splitting ( @xmath467 ) is negligible @xcite , i.e. their evolution can be treated with balance equations ( equations for state populations ) .
coherence effects prove to be important only for four nearly degenerate pairs of states , of type of the tls discussed above , which describe resonances non - overlapping in the considered limit .
these four pairs can be combined into two groups of pairs of these tlss : @xmath468 and @xmath469 , respectively .
transitions in tls - pairs @xmath470 are associated with those in corresponding separate radical @xmath242 .
tls - states @xmath471 , corresponding to the zeroth z - projection of the total spin ( @xmath472 ) , are the same for systems @xmath473 and @xmath474 . however
, these systems can be considered as uncoupled because in the studied limit of large @xmath475 significantly efficient transitions in systems @xmath237 and @xmath238 occur at different values of @xmath8 ( i.e. corresponding resonances do not overlap , as it was mentioned above ) . for
this reason it is possible to distinguish the same states @xmath476 , belonging to @xmath237- and @xmath238-systems , and denote them as @xmath477 or @xmath478 , respectively ( the subscript @xmath479 or @xmath480 can be omitted as it will be explained below ) .
the assumed initial population of the singlet state reduces to that of the above - mentioned states @xmath477 and @xmath477 ( in which @xmath472 ) with the probability @xmath481 .
notice that these states are reactive .
the reactivity matrices @xmath482 are similar for all systems and quite accurately determined as the two level variant of formula ( [ mfe4 ] ) .
these matrices describe reaction with the same rate approximately equal to @xmath483 , where @xmath484 is the reactivity in @xmath247-state .
all the tlss give the same contribution to the total yield @xmath7 , differing only in the resonance frequency ( @xmath262 or @xmath263 ) if they correspond to different radicals .
therefore we can combine the identical contributions of the tlss @xmath485 and @xmath486 into one @xmath487 of two times larger magnitude and omit subscripts @xmath488 and @xmath489 in the notation of the tlss and their parameters , as it has been mentioned above . in so doing we arrive at the representation of @xmath7 in the form @xmath490 where @xmath491 are given by @xmath492 \label{mfe6}\ ] ] with @xmath493 and @xmath494 . for simplicity
, we also assume that the microwave field @xmath9 is weak enough so that the effects of the @xmath9-induced transitions can be treated perturbatively in the lowest order expansion of @xmath7 in @xmath495 . in these assumptions
the yield @xmath487 can be evaluated with approximate expression @xmath496
\label{mfe7}\ ] ] in which @xmath455 and @xmath497 .
calculation using eq .
( [ mfe6 ] ) gives for the magnetic field dependent part @xmath498 , which is called rydmr spectrum , @xmath499 where @xmath500 is written as @xmath501 \label{mfe10a } \\ & = & -\frac{\cos ( \pi \alpha /2)}{2}\bigg(\frac{l_s}{l_{\mu}}\bigg)^{\!2}\!\ !
\frac{(\omega_1/w)^{\alpha}}{(1 + \omega^2/\omega_1 ^ 2)^{1-\alpha/2}}.\qquad \label{mfe10b}\end{aligned}\ ] ] it is worth noting some important specific features of the rydmr spectrum ( in the case of anomalous diffusion ) predicted by eq .
( [ mfe10b ] ) : \1 ) unlike conventional markovian migration , anomalous diffusion leads to the non - analytical dependence of the spectrum on @xmath502 ( i.e. on the parameters of spin hamiltonian ) which can be obtained in the lowest order in @xmath502 .
naturally , as in the case of mary , the lowest - order value of rydmr vanishes in the limit @xmath290 because at @xmath287 , corresponding to conventional caging , rydmr amplitude is determined by the second order term of expansion in @xmath502 .
\2 ) at large @xmath8 ( at line wings ) the rydmr resonance contributions @xmath503 decrease as @xmath504 , i.e. slower than the lorenzian line ( @xmath505 ) .
\3 ) the width of resonances in the spectrum is determined by the amplitude of microwave field @xmath9 . in other words
these kind of spectra are always measured in the saturation regime @xcite .
\4 ) at first sight , the fact that the width is determined by @xmath9 is a consequence of long memory effects on the processes governed by anomalous diffusion , i.e the absence of the characteristic time in these processes .
therefore in the presence of such time caused , for example , by the conventional intraradical spin lattice relaxation , the width seems to depend on this time . in reality , however , this is not true which can easily be demonstrated in a simple model assuming this time to result from the spin independent decay of radicals with the rate @xmath506 . in this model
the magnetic field dependent yield contributions @xmath507 are still given by eq .
( [ mfe10a ] ) but with @xmath508 replaced by @xmath509^{\alpha } - ( w_0/w)^{\alpha}$ ] : @xmath510 . \label{mfe11}\ ] ] it is easily seen that in the limit @xmath511 this expression reproduces eq .
( [ mfe10b ] ) while in the opposite limit it predicts @xmath512
. therefore the presence of the characteristic relaxation time does not lead to the change of the line width of rydmr spectra slightly changing only the line shape .
this work concerns the detailed analysis of the specific features of relaxation kinetics in quantum systems induced by anomalous noise .
two types of quantum systems are considered , as examples : two - level systems and radical pairs in a potential well whose recombination is assisted by anomalous diffusion .
the analysis is made with the use of the recently developed convenient and powerful method based on the ctrwa and the non - markovian sle .
it demonstrated some important peculiarities of the kinetics of the processes under study .
first of all , the relaxation kinetics in both type of systems proved to be strongly non - exponential .
more subtle peculiarities were found in spectral characteristics of these processes : the line shape , its dependence on the parameters of processes , etc . in this work
we mainly restricted ourselves to analytical analysis of the processes , however , as it was point out above , the proposed method also allows for significant simplification of numerical treatment of the processes under study , especially in much more complicated multilevel quantum systems : magnetic clusters , magnetic glasses , etc .
1 : the spectrum @xmath513 , where @xmath514 , calculated in the model ( [ two3 ] ) [ using eq . ( [ res4c ] ) ] for two values of @xmath304 : @xmath515 ( a ) and @xmath516 ( b ) , and different values of @xmath6 : @xmath517 ( full ) , ( 2 ) @xmath518 ( dashes ) , and @xmath519 ( dots ) . fig .
3 : the spectrum @xmath513 , where @xmath522 , calculated in the model ( [ two3 ] ) [ using eq . ( [ res4c ] ) ] for two values of @xmath6 : @xmath523 ( a ) and @xmath524 ( b ) , and different values of @xmath525 : @xmath526 ( dots ) , @xmath527 ( dashes ) , @xmath528 ( full ) and @xmath529 ( dash - dots ) . fig .
4 : population relaxation kinetics @xmath530 , where @xmath531 , calculated with eq .
( [ res7 ] ) ( full lines ) for two values of @xmath525 : @xmath532 ( a ) and @xmath533 ( b ) , and different values of @xmath6 : ( 1 ) @xmath534 , ( 2 ) @xmath519 , ( 3 ) @xmath535 , and ( 4 ) @xmath536 .
straight ( dashed ) lines represent exponential dependence [ eq .
( [ res6 ] ) ] . | the kinetics of relaxation in quantum systems induced by anomalously slowly fluctuating noise is studied in detail . in this study
two processes are considered , as examples : ( 1 ) relaxation in two - level system ( tls ) caused by external noise with slowly decaying correlation function @xmath0 , where @xmath1 , and ( 2 ) anomalous - diffusion controlled radical pair ( rp ) recombination in which relaxation results from the reaction with slowly fluctuating reaction rate whose fluctuations are governed by subdiffusive relative motion of radicals in a potential well .
analysis of these two processes is made within continuous time random walk approach ( ctrwa ) .
rigorous ctrwa - treatment of the processes under study results in the non - markovian stochastic liouville equation ( sle ) for the density matrix of the systems .
this sle predicts important specific features of relaxation kinetics of quantum systems in the presence of the above mentioned anomalous noise . in tls ,
for example , relaxation of both phase and population turns out to be anomalously slow and strongly non - exponential .
moreover , for @xmath2 in the limit of characteristic fluctuation rates @xmath3 much larger than the frequency of quantum transitions @xmath4 ( @xmath5 ) the relaxation kinetics in tls is independent of @xmath3 .
strong changes of the relaxation kinetics is found with the change of @xmath3 and @xmath6 . in rp recombination
the anomalous fluctuations of reaction rate ( due to anomalous diffusion ) show themselves in non - analytical dependence of the reaction yield @xmath7 on reactivity and parameters of the rp spin hamiltonian .
in particular , the spectrum shape of reaction yield detected magnetic resonance , i.e. the dependence of @xmath7 on the frequency @xmath8 of resonance microwave field is found to be strongly non - lorenzian with the width to be determined by the strength @xmath9 of this field . |
the energetic gamma - ray experiment telescope ( egret ) on - board the compton gamma - ray observatory ( cgro ; 1991@xmath82000 ) was the first to detect high - energy emission from grbs .
egret detected only five grbs with its spark chambers ( 20@xmath9mev to 30@xmath9gev ) and a few grbs with its total absorption shower counter ( tasc ; @xmath10mev ) .
nevertheless , these events already showed diversity .
the most prominent examples are grb 940217 , with high - energy emission lasting up to @xmath11hr after the grb including an @xmath12gev photon after @xmath13hr , @xcite and grb 941017 which had a distinct high - energy spectral component @xcite detected up to @xmath14mev with @xmath15 .
this high - energy spectral component had @xmath16 times more energy and lasted longer ( @xmath14s ) than the low - energy ( hard x - ray to soft gamma - ray ) spectral component ( which lasted several tens of seconds ) , and may be naturally explained as inverse - compton emission from the forward - reverse shock system that is formed as the ultra - relativistic grb outflow is decelerated by the external medium @xcite .
nevertheless , better data are needed in order to determine the origin of such high - energy spectral components more conclusively .
the italian experiment astro - rivelatore gamma a immagini leggero ( agile ; launched in 2007 ) has detected grb 080514b at energies up to @xmath17mev , and the high - energy emission lasted longer ( @xmath213@xmath1s ) than the low - energy emission ( @xmath18s ) @xcite .
below are some highlights of fermi grb observations so far and what they have taught us .
during its first @xmath19yr of routine operation , from aug . 2008 to jan .
2010 , the lat has detected 14 grbs , corresponding to a detection rate of @xmath20 .
table [ tab : lat14grbs ] summarizes their main properties . while at least 13 of the 14 lat grbs had @xmath21 photons above @xmath0mev , 4 were particularly bright in the lat with @xmath22 photon above @xmath23gev , @xmath21 photons above @xmath24gev , and @xmath25 photons above @xmath0mev .
this corresponds to a bright lat grb ( as defined above ) detection rate of @xmath26grb / yr ( with a rather large uncertainty due to the small number statistics ) .
there were also 11 grbs with @xmath22 photon above @xmath24gev , corresponding to @xmath27grb / yr . these detection rates are compatible with pre - launch expectations @xcite based on a sample of bright batse grbs for which the fit to a band spectrum over the batse energy range ( @xmath28kev to @xmath29mev ) was extrapolated into the lat energy range ( see fig .
[ fig : pre - launch ] ) .
the agreement is slightly better when excluding cases with a rising @xmath30 spectrum at high energies ( i.e. a high - energy photon index @xmath31 ) .
] this suggests that , on average , there is no significant excess or deficit of high - energy emission in the lat energy range relative to such an extrapolation from lower energies . as described in
[ sec : delay+hecomp ] , however , in individual cases we do have evidence for such an excess .
the observed lat grb detection rate implies that , on average , only about @xmath32 of the energy that is radiated during the prompt grb emission phase is channeled into the lat energy range , suggesting that in most grbs the high - energy radiative output does not significantly affect the total energy budget .
short grbs , however , appear to be different in this respect ( see [ sec : long - vs - short ] and fig . [
fig : long - vs - short ] ) .
gev , respectively ) , while their hight indicates the uncertainty ( @xmath33yr ) on the corresponding lat detection rate ( @xmath34yr ) due to the small number ( @xmath35 ) of detected grbs.,scaledwidth=92.0% ] grb & @xmath36 & or & & starts & lasts & spec . & energy & @xmath37 & & short & 0.1@xmath9gev & 1@xmath9gev & later & longer & comp . & ( gev ) & + 080825c & @xmath3860@xmath39 & long & @xmath3810 & 0 & ? & yes & no & @xmath380.6 & + 080916c & @xmath40 & long & 145 & 14 & yes & yes & ? & @xmath3813 & @xmath384.35 + 081024b & @xmath41 & short & @xmath3810 & 2 & yes & yes & ? & @xmath383 &
+ 081215a & @xmath3886@xmath39 & long & & & ? & ? &
& & + 090217 & @xmath3834@xmath39 & long & @xmath3810 & 0 & no & no & no & @xmath381 & + 090323 & @xmath3855@xmath39 & long & @xmath3820 & @xmath20 & ?
& yes & ? & ? & 3.57 + 090328 & @xmath3864@xmath39 & long & @xmath3820 & @xmath20 & ? & yes & ? & ? & 0.736 + 090510 & @xmath3814@xmath39 & short & @xmath2150 & @xmath220 & yes & yes & yes & @xmath3831 & 0.903 + 090626 & @xmath3815@xmath39 & long & @xmath3820 & @xmath20 & ? & yes & ? & ? &
+ 090902b & @xmath42 & long & @xmath2200 & @xmath230 & yes & yes & yes & @xmath3833 & 1.822 + 090926 & @xmath3852@xmath39 & long & @xmath2150 & @xmath250 & yes & yes & yes & @xmath3820 & 2.1062 + 091003a & @xmath3813@xmath39 & long & @xmath3820 & @xmath20 & ? & ? & ? & ? &
0.8969 + 091031 & @xmath3822@xmath39 & long & @xmath3820 & @xmath20 & ? & ? & ? & @xmath381.2 &
+ 100116a & @xmath3829@xmath39 & long & @xmath3810 & 3 & ? & ? & ? & @xmath382.2 &
some quantum gravity models allow violation of lorentz invariance , and in particular allow the photon propagation speed , @xmath43 , to depend on its energy , @xmath44 : @xmath45 , where @xmath46 .
the lorentz invariance violating ( liv ) part in the dependence of the photon momentum , @xmath47 , on its energy , @xmath44 , can be expressed as a power series , @xmath48 in the ratio of @xmath44 and a typical energy scale @xmath49 for the @xmath50 order , which is expected to be up to the order of the planck scale , @xmath51 , where @xmath52 . since we observe photons of energy well below the planck scale ,
the dominant liv term is associated with the lowest order non - zero term in the sum , of order @xmath53 , which is usually assumed to be either linear ( @xmath54 ) or quadratic ( @xmath55 ) .
the photon propagation speed is given by the corresponding group velocity , @xmath56\ .\ ] ] note that @xmath57 corresponds to the sub - luminal case ( @xmath58 and a positive time delay ) , while @xmath59 corresponds to the super - luminal case ( @xmath60 and a negative time delay ) .
taking into account cosmological effects @xcite , this induces a time delay ( or lag ) in the arrival of a high - energy photon of energy @xmath61 , compared to a low - energy photon of energy @xmath62 ( emitted simultaneously at the same location ) , of @xmath63 here we concentrate on our results for a linear energy dependence ( @xmath64 ) .
we have applied this formula to the highest energy photon detected in grb 080916c , with an energy of @xmath65gev , which arrived at @xmath66s after the grb trigger ( i.e. the onset of the @xmath67mev emission ) .
since it is hard to associate the highest energy photon with a particular spike in the low - energy lightcurve , we have made the conservative assumption that it was emitted anytime after the grb trigger , i.e. @xmath68 , in order to obtain a limit for the sub - luminal case ( @xmath57 ) : @xmath69 .
this was the strictest limit of its kind @xcite , at that time .
however , the next very bright lat grb , 090510 , was short and had very narrow sharp spikes in its light curve ( see fig [ fig:090510-liv ] ) , thus enabling us to do even better @xcite . our main results for grb 090510 are summarized in table [ tab : liv ] .
the first 4 limits are based on a similar method as described above for grb 080916c , using the highest energy photon , @xmath70gev , and assuming that its emission time @xmath71 was after the start of a relevant lower energy emission episode : @xmath72 .
these 4 limits correspond to different choices of @xmath73 , which are shown by the vertical lines in fig .
[ fig:090510-liv ] .
we conservatively used the low end of the @xmath74 confidence interval for the highest energy photon ( @xmath75gev ) and for the redshift ( @xmath76 ) .
the most conservative assumption of this type is associating @xmath73 with the onset of any detectable emission from grb 090510 , namely the start of the small precursor that gbm triggered on , leading to @xmath77 .
however , it is highly unlikely that the @xmath78gev photon is indeed associated with the small precursor .
it is much more likely associated with the main soft gamma - ray emission , leading to @xmath79 . moreover , for any reasonable emission spectrum , the emission of the @xmath78gev photon would be accompanied by the emission of a large number of lower energy photons , which would suffer a much smaller time delay due to liv effects , and would therefore mark its emission time .
we could easily detect such photons in energies above @xmath0mev , and therefore the fact that significant high - energy emission is observed only at later times ( see fig . [ fig:090510-liv ] ) strongly argues that the @xmath78gev photon was not emitted before the onset of the observed high - energy emission .
one could choose either the onset time of the emission above @xmath0mev or above @xmath24gev , which correspond to @xmath80 , and @xmath81 , respectively .
gev photon was emitted near our choices for @xmath82 , together with lower energy photons , as can be expected for any reasonable emission spectrum .
this is evident from the lack of accumulation of photons along the _ solid _ curves in panel ( a ) of fig .
[ fig:090510-liv ] , at least for the first 3 @xmath73 values , and provides support for these choices of @xmath73 ( i.e. that they can indeed serve as upper limits on a liv induced energy dispersion ) . ] the 5@xmath83 and 6@xmath83 limits in table [ tab : liv ] are more speculative , as they rely on the association of an individual high - energy photon with a particular spike in the low - energy light curve , on top of which it arrives . while these associations are not very secure ( the chance probability is roughly @xmath84 ) , they are still most likely , making the corresponding limits interesting , while keeping this big caveat in mind .
the allowed emission time of these two high - energy photons , if these associations are real , is shown by the two thin vertical shaded regions in fig . [ fig:090510-liv ] . for the @xmath78gev photon
this gives a limit of @xmath85 for either sign of @xmath86 .
the last limit in table [ tab : liv ] is based on a different method , which is complementary and constrains both signs of @xmath86 .
it relies on the highly variable high - energy light curve , with sharp narrow spikes , which would be smeared out if there was too much energy dispersion ( of either sign ) .
we have used the discan method @xcite to search for linear energy dispersion within the lat data ( the actual energy range of the photons used was @xmath87mev to @xmath78gev ) gev or @xmath24gev . ] during the most intense emission interval ( 0.5@[email protected]@xmath9s ) .
this approach extracts dispersion information from all detected lat photons and does not involve binning in time or energy . using this method we obtained a robust lower limit of @xmath88 ( at the 99% confidence level ) .
our most conservative limits ( the first and last limits in table [ tab : liv ] ) rely on very different and largely independent analysis , yet still give a very similar limit : @xmath89 .
this lends considerable support to this result , and makes it more robust and secure than for each of the methods separately .
.lower - limits on the quantum gravity ( qg ) mass scale associated with a possible linear ( @xmath90 ) variation of the speed of light with photon energy , that we can place from the lack of time delay ( of sign @xmath86 ) in the arrival of high - energy photons relative to low - energy photons , from our observations of grb 090510 ( from @xcite ) . [ tab : liv ] [ cols="^,^,^,^,^,^ " , ]
the grb prompt emission typically has very large isotropic equivalent luminosities ( @xmath91 ) , significant short time scale variability , and typical photon energies @xmath92 ( in the source cosmological frame ) .
this would result in a huge optical depth to pair production ( @xmath93 ) at the source , which would thermalize the spectrum and thus be at odds with the observed non - thermal spectrum , unless the emitting material was moving toward us relativistically , with a bulk lorentz factor @xmath94 .
this `` compactness '' argument has been used to derive lower - limits , @xmath95 , on the value of @xmath96 , which were typically @xmath97 and in some cases as high as a few hundred ( see @xcite and references therein ) .
however , the photons that provided the opacity for these limits were well above the observed energy range , so there was no direct evidence that they actually existed in the first place . with fermi
, however , we adopt a more conservative approach of relying only on photons within the observed energy range . under this approach ,
@xmath98 where @xmath99 is the highest observed photon energy , so that setting a large @xmath95 requires observing sufficiently high - energy photons .
the main uncertainty in deriving @xmath95 is usually the exact choice for the variability time , @xmath100 .
other uncertainties arise from those on the spectral fit parameters , or on the degree of space - time overlap between the high - energy photon and lower energy target photons , in cases where there is more than one spectral component without conclusive temporal correlation between their respective light curves .
finally , the fact that our limits rely on a single high - energy photon also induces an uncertainty , as it might still escape from an optical depth of up to a few .
however , in most cases the second or third highest - energy photons help to relax the affect this has on @xmath95 ( as the probability that multiple photons escape from @xmath101 rapidly decreases with the number of photons ) .
thus , we have derived reasonably conservative @xmath95 values for 3 of the brightest lat grbs : @xmath102 for grb 080916c @xcite , @xmath103 for grb 090510 @xcite , and @xmath104 for grb 090902b @xcite .
this shows that short grbs ( such as 090510 ) are as highly relativistic as long grbs ( such as 080916c or 090902b ) , which was questioned before the launch of fermi @xcite . since our highest values of @xmath95
are derived for the brightest lat grbs , they are susceptible to strong selection effects
. it might be that grbs with higher @xmath96 tend to be brighter in the lat energy range ( e.g. by avoiding intrinsic pair production @xcite ) .
a delayed onset of the high - energy emission ( @xmath2100@xmath1mev ) relative to the low - energy emission ( @xmath105mev ) appears to be a very common feature in lat grbs .
it clearly appears in all 4 of the particularly bright lat grbs , while in dimmer lat grbs it is often inconclusive due to poor photon statistics near the onset time .
the time delay , @xmath106 , appears to scale with the duration of the grb ( @xmath107 several seconds in the long grbs 080916c and 090902b , while @xmath108s in the short grbs 090510 and 081024b , though with a smaller significance for the latter ) .
only 3 lat grbs so far have shown clear ( @xmath25@xmath109 ) evidence for a distinct spectral component . however , these grbs are the 3 brightest in the lat , while the next brightest grb in the lat ( 080916c ) showed a hint for an excess at high energies .
this suggests that such a distinct high - energy spectral component is probably very common , but we can clearly detect it with high significance only in particularly bright lat grbs , since a large number of lat photons is needed in order to detect it with @xmath25@xmath109 significance .
the distinct spectral component is usually well fit by a hard power - law that dominates at high energies . in grb
090902b a single power - law component dominates over the usual band component both at high energies ( above @xmath110mev ) and low energies ( below @xmath111kev ; see _ lower panel _ of fig .
[ fig : long - vs - short ] ) .
there is also marginal evidence that the high - energy power - law component in grb 090510 , which dominates above @xmath110mev , might also appear at the lowest energies ( below a few tens of kev ) .
both the delayed onset and distinct spectral component relate to and may help elucidate the uncertain prompt grb emission mechanism .
the main two competing classes of models are leptonic and hadronic origin . * leptonic : * the high - energy spectral component might be inverse - compton emission , and in particular synchrotron - self compton ( ssc ) if the usual band component is synchrotron . in this case
, however , it may be hard to produce the observed @xmath112 , where @xmath100 is the width of individual spikes in the lightcurve ( @xmath113 might occur due to the build - up of the seed synchrotron photon field in the emitting region over the dynamical time ) .
moreover , the gradual increase in the photon index @xmath114 of the distinct high - energy power - law spectral component is not naturally expected in such a model , and the fact that it is different than the band low - energy photon index as well as the excess flux ( above the band component ) at low energies are hard to account for in this type of model . *
hadronic : * @xmath106 might be identified with the acceleration time , @xmath115 , of protons ( or heavier ions ) to very high energies ( at which they loose much of their energy on a dynamical time , e.g. via proton synchrotron @xcite , in order to have a reasonable radiative efficiency ) .
if the observed high - energy emission ( and in particular the distinct high - energy spectral component ) also involves pair cascades ( e.g. inverse - compton emission by secondary @xmath116 pairs @xcite produced in cascades initiated by photo - hadronic interactions ) then it might take some additional time for such cascades to develop . such an origin for @xmath106 ( @xmath117 )
, however , requires the high - energy emission to originate from the same physical region over times @xmath118 , and implies high - energy emission rise and variability times @xmath119 , due to the stochastic nature of the acceleration process ( while @xmath120 is usually observed ) .
the gradual increase in @xmath114 is not naturally expected in hadronic models , though it might be mimicked by a time - evolution of a high - energy band - like spectral component @xcite .
for grb 090510 a hadronic model requires a total isotropic equivalent energy @xmath121 times larger than that observed in gamma - rays @xcite , which may pose a serious challenge for the progenitor of this short grb .
the excess flux at low energies that is observed in grb 090902b ( and the hint for such an excess in grb 090510 ) may be naturally explain in this type of model by synchrotron emission from secondary pairs @xcite .
altogether , hadronic models seem to fare somewhat better , however both leptonic and hadronic models still face many challenges , and do not yet naturally account for all of the fermi observations .
in most lat grbs the high - energy ( @xmath2100@xmath1mev ) emission lasts significantly longer than the low - energy ( @xmath105mev ) emission . while the high - energy emission usually shows significant variability during the prompt ( low - energy ) emission phase , in some cases showing temporal correlation with the low - energy emission , the longer lived emission is typically temporally smooth and consistent with a power - law flux decay ( of @xmath122 ) with a lat photon index corresponding to a roughly flat @xmath123 .
it is most natural to interpret the prompt high - energy emission as the high - energy counterpart of the prompt soft gamma - ray emission , from the same emission region , especially when there is temporal correlation between the low and high - energy light curves , and sometimes even from the same spectral component ( as appears to be the case for grb 080916c ) .
however , when there is no such temporal correlation , an origin from a different emission region is possible .
the longer lived smooth power - law decay phase is more naturally attributed to the high - energy afterglow , from the forward shock that is driven into the external medium .
an afterglow origin has been suggested in some cases for the whole lat emission @xcite , including during the prompt soft gamma - ray emission stage , however in this scenario it is generally hard to explain the sharp spikes in the lat lightcurve during the prompt phase .
it is easier to test the origin of this long lived high - energy emission when there is good multi - wavelength coverage of the early afterglow emission ( e.g. , in x - ray and/or optical ) , such as for grb 090510 @xcite .
producing particularly high - energy photons is challenging for a synchrotron origin , both during the prompt emission @xcite , and even more so during the afterglow ( e.g. @xcite ) , as it requires a very high bulk lorentz factor and upstream magnetic field , in addition to a very efficient shock acceleration ( e.g. a @xmath124gev photon observed in grb 090902b after @xmath125s , well after the end of the prompt emission @xcite , requires @xmath126 ) .
@xmath910@xmath1gev ) versus low ( 20@xmath9kev @xmath127mev ) energies ( from @xcite ) , for 4 long ( 080825c , 080916c , 090217 , 090902b ) and 2 short ( 081024b , 090510 ) duration lat grbs .
the diagonal lines indicate high to low energy fluence ratios of 1% , 10% , and 100% . *
bottom panel * : the best fit time - integrated @xmath123 spectra for the same grbs , two of which ( 090510 , 090902b ) show a distinct spectral component , well described by a hard power - law , in addition to the usual band spectral component .
the colored shaded regions indicate the energy ranges used for calculating the fluences that are displayed in the _
top panel_.,scaledwidth=70.0% ] @xmath910@xmath1gev ) versus low ( 20@xmath9kev @xmath127mev ) energies ( from @xcite ) , for 4 long ( 080825c , 080916c , 090217 , 090902b ) and 2 short ( 081024b , 090510 ) duration lat grbs .
the diagonal lines indicate high to low energy fluence ratios of 1% , 10% , and 100% . *
bottom panel * : the best fit time - integrated @xmath123 spectra for the same grbs , two of which ( 090510 , 090902b ) show a distinct spectral component , well described by a hard power - law , in addition to the usual band spectral component .
the colored shaded regions indicate the energy ranges used for calculating the fluences that are displayed in the
_ top panel_.,scaledwidth=70.0% ]
so far , 2 ( 12 ) out of the 14 lat grbs are of the short ( long ) duration class .
this implies that @xmath128 of lat grbs are short , with a large uncertainty due to the small number statistics , which is consistent with the @xmath129 short grbs detected by the gbm . as can be seen from table [ tab : lat14grbs ] , the high - energy emission properties of short and long grbs appear to be rather similar .
they can both produce very bright emission in the lat energy range ( 090510 vs. 080916c , 090902b and 090926 ) , with a correspondingly high lower - limit on the bulk lorentz factor ( @xmath130 ) , as well a distinct spectral component ( 090510 vs. 090902b and 090926 ) .
both show a delayed onset and loner lived high - energy emission , compared to the low - energy emission
. however , the delay in the onset of the high - energy emission appears to roughly scale with the duration of the grb , being @xmath131s for short grbs and several seconds for long grbs .
this is especially intriguing when comparing grbs 080916c and 090510 , which had a comparable isotropic equivalent luminosity ( of several @xmath132 ) , suggesting another underlying cause for the difference in the time delay ( e.g. @xcite ) . another interesting potential difference , which still needs to be confirmed ( as there are only 2 short lat grbs so far , and possible selection effects ) , is that short grbs appear to have a comparable energy output at high and low photon energies , while long grbs tend to radiate a smaller fraction of their energy output at high photon energies ( see _ upper panel _ of fig .
[ fig : long - vs - short ] ) . | the fermi gamma - ray space telescope has more than doubled the number of gamma - ray bursts ( grbs ) detected above @xmath0mev within its first year of operation .
thanks to the very wide energy range covered by fermi s gamma - ray burst monitor ( gbm ; 8@xmath1kev to 40@xmath1mev ) and large area telescope ( lat ; 25@xmath1mev to @xmath2300@xmath1gev ) it has measured the prompt grb emission spectrum over an unprecedentedly large energy range ( from @xmath3kev to @xmath4gev ) .
here i briefly outline some highlights from fermi grb observations during its first @xmath5yr of operation , focusing on the prompt emission phase .
interesting new observations are discussed along with some of their possible implications , including : ( i ) what can we learn from the fermi - lat grb detection rate , ( ii ) a limit on the variation of the speed of light with photon energy ( for the first time beyond the planck scale for a linear energy dependence from direct time of arrival measurements ) , ( iii ) lower - limits on the bulk lorentz factor of the grb outflow ( of @xmath6 for the brightest fermi lat grbs ) , ( iv ) the detection ( or in other cases , lack thereof ) of a distinct spectral component at high ( and sometimes also at low ) energies , and possible implications for the prompt grb emission mechanism , ( v ) the later onset ( and longer duration ) of the high - energy emission ( @xmath2100@xmath1mev ) , compared to the low - energy ( @xmath71@xmath1mev ) emission , that is seen in most fermi - lat grbs . |
while this paper was drafted to convey a result of the author s the first half of this paper is devoted to the presentation of harrington s results from @xcite as the technique he used to settle mclaughlin s conjecture is needed for the author s result and has never before appeared in print .
the author would like to make absolutely clear that these results are harrington s alone , but as harrington s notes are quite sparse and the way in which he anticipated filling in the details has faded with time , the details are , for good or ill , the author s own .
in addition to various assorted details the technical results in [ sec : ord - org ] on nice ordinal notations and the modifications required to prove lemma [ lem : subgeneric - root ] and corollary [ cor : subgeneric - star ] true are of the author s devising and it is unclear what it any resemblance they might bear to harrington s original conception of these proofs . once harrington s method has been presented the second half of the paper will revert to a more standard style and provide a brief review of previously published literature on fast growing functions and turing degrees followed by the author s own results in this area . the notation we use is largely standard .
we use @xmath4 to denote the integer code of the pair @xmath5 , @xmath6 for the set whose @xmath7-th column is @xmath8 , and @xmath9 to denote the compliment of @xmath10 . a string is a member of @xmath11 and trees subsets of @xmath11 closed under initial segments . when we need to distinguish between strings and their integer codes we write @xmath12 for the code of @xmath13 .
we use @xmath14 and @xmath15 to denote that @xmath16 are incompatible and compatible respectively and write @xmath17 to denote the concatenation of the two strings .
@xmath18 gives the length of @xmath19 and @xmath20 denotes the longest proper initial segment of @xmath19 .
the set of ( infinite ) paths through a tree @xmath21 is denoted @xmath22 $ ] and @xmath23 is the set of strings in @xmath21 extended by some infinite path .
we call functions from @xmath11 to @xmath11 monotonic if it is an isomorphism of the partial ordering @xmath24 on it s domain and range .
we abuse notation and use @xmath25 to denote the members of @xmath21 of length at most @xmath26 and write @xmath27 for @xmath28 when @xmath29 is monotonic and total on @xmath30 .
kleene s set of ordinal notations is @xmath31 , the canonical ordering of notations is @xmath32 and @xmath33 gives the effective sum of notations .
when @xmath34 is a limit notation we denote the @xmath26-th element of the effectively given increasing sequence defining @xmath34 by @xmath35 .
we write @xmath36 and @xmath37 for the collections of computably @xmath38 and @xmath39 formulas and @xmath40{\alpha } $ ] and @xmath41{\alpha } $ ] when a predicate for membership in @xmath42 is introduced into the language .
we use @xmath43 and @xmath44 to denote infinite disjunction and infinite conjunction respectively .
we refer the reader to @xcite for more on computable infinitary formula and to @xcite for more on @xmath31 .
we do introduce a few non - standard pieces of notation particular to the subject matter .
given partial functions @xmath45 we write @xmath46 if @xmath47 whenever they are both defined .
when @xmath48 and @xmath49 are total functions we read @xmath50 as @xmath51 majorizes @xmath52 .
we say @xmath51 dominates @xmath52 if some @xmath53 differing from @xmath51 at finitely many locations majorizes @xmath52 .
we indicate the local forcing relation on @xmath23 by @xmath54 $ ] and it s relativization to @xmath55 by @xmath56 $ ] and refer the reader to @xcite for the definitions of the standard forcing relation @xmath57 and @xcite for local forcing .
informally , @xmath54 $ ] is defined in the same manner as @xmath57 except with all quantifications over @xmath11 replaced with quantification s over @xmath23 ( nodes in @xmath21 that extend to paths ) .
when we extend the usual language of forcing by introducing a predicate symbol for membership in @xmath42 we write @xmath58 \phi $ ] to indicate that @xmath59 can check membership in @xmath42 as an atomic operation .
usually the set @xmath42 we are forcing relative to will be clear from context and we will simply write @xmath60 \phi $ ] .
when @xmath61 \phi $ ] or @xmath61 \lnot\phi $ ] for every @xmath59 in @xmath40{\beta } $ ] we say that @xmath52 is @xmath62 generic on @xmath21 relative to @xmath42 .
we will take our forcing relation to denote strong forcing , that is @xmath13 forces @xmath63{1 } $ ] sentences only when @xmath64 , i.e. , @xmath59 is satisfied by referring only to information in @xmath19 .
it is important to note that our notion of @xmath51 being @xmath2 generic on @xmath21 does not require @xmath51 to force all @xmath65{\alpha } $ ] facts or their negations as some definitions of genericity on a tree require @xcite but only @xmath36 facts nor does it require that @xmath51 be non - isolated .
our definition is the natural way to preserve
the notion of a generic path as one on which every truth is determined by a finite initial segment while requiring @xmath2 generic paths on @xmath21 to force all @xmath65{\alpha } $ ] facts or their negations extends the idea that a generic path should be typical .
thus under our definition there is a perfect tree @xmath21 with every path through @xmath21 @xmath2 generic on @xmath21 while this would be impossible under the other notion . while our standard notion of forcing is concerned only with the extendable nodes on @xmath21 we will also make use of a more effective notion that , by analogy with the notion of strong forcing , we call super forcing on @xmath21 denoted @xmath66 $ ]
. the definition of super forcing on @xmath21 exactly mirrors the definition in @xcite of strong forcing modified as usual to get the local forcing relation on @xmath21 instead of @xmath23 as above .
that is for @xmath19 to force @xmath67 on @xmath21 requires that every @xmath68 with @xmath69 satisfy @xmath70 \phi $ ] while for @xmath19 to strongly force @xmath67 on @xmath21 requires this hold for every @xmath68 with @xmath71 .
hence @xmath72[\zeron{\beta } ] \phi $ ] for @xmath63{\alpha } $ ] is @xmath73{\alpha } $ ] and @xmath74{\alpha } $ ] for @xmath75{\alpha } $ ] .
in @xcite harrington answered mclaughlin s conjecture in the negative and we will adapt his construction to establish theorem [ thm : nonu ] but we first present his approach . while other variations on the theme have been called mclaughlin s conjecture the form of the conjecture refuted by harrington in @xcite is the one appearing in @xcite that asserts : [ conj : mclaughlin ] every element of a countable arithmetic subset of @xmath76 is an arithmetic singleton . harrington s refutation consisted of the following theorem .
[ thm : harrington - mcl ] for every computable tree @xmath77 is a computable tree @xmath78 such @xmath22 $ ] and @xmath79 $ ] are homeomorphic and every @xmath80 $ ] is arithmetically ( @xmath81 ) generic on @xmath22 $ ] .
[ cor : mcl - false ] is false .
we sketch how harrington s result contradicts mclaughlin s conjecture .
let @xmath82 have some non - isolated path @xmath83 .
thus the homeomorphic image of @xmath83 , @xmath51 is a non - isolated path through @xmath21 .
now suppose that @xmath84 is an arithmetic predicate with unique solution @xmath51 .
by genericity we must have @xmath85 \phi $ ] hence some @xmath86 forces @xmath59 . as @xmath51 non - isolated
there is some @xmath87 also extending @xmath19 . at @xmath88
is also @xmath81 generic on @xmath21 and @xmath89 \phi $ ] we have @xmath90 .
contradiction .
once we know that harrington s result is true the natural approach for a recursion theorist is simply to go out and build @xmath21 as some kind of distorted copy of @xmath82 while trying to meet the genericity requirements . the natural approach would be to simply go ahead and try to build @xmath21 directly but of course if that worked straightforwardly the conjecture would likely never have remained open for as it did .
in particular the ` nested ' nature of the genericity requirements makes direct construction of @xmath21 extremely difficult . to force @xmath92 facts about @xmath51 we need to react to the _ particular _ way we ve failed to force @xmath93 facts about @xmath51 .
were we building @xmath51 to be fully generic there would be no question about how @xmath51 forced @xmath93 and @xmath94 facts .
we could simply read off from the definition of forcing whether a given @xmath86 forced some instance @xmath95 of a @xmath94 formula and simply require @xmath51 to extend some appropriate @xmath96 .
but as we clearly ca nt build our desired @xmath51 to be even fully @xmath97-generic here we must sometimes bring it about that @xmath86 forces some @xmath94 formula @xmath98 despite the fact that @xmath59 is true on a co - meager set in @xmath76 by pruning from @xmath21 all extensions of @xmath19 that force @xmath59 . doing this on it s own while keeping @xmath21 computable
would be organizationally difficult but if we are to keep @xmath22=[t_\omega ] $ ] we must somehow also anticipate when our commitment to somehow copy @xmath99 will be incompatible with trying to force a sentence in a particular direction . therefore , rather than a frontal assault harrington described how we can attack the problem in reverse in a manner that provides all our organization for free .
the approach taken by harrington rests on projecting down @xmath99 to a sequence on intermediate trees @xmath100 with @xmath101 .
each @xmath100 will be computable in @xmath102 and @xmath103 will be the image of @xmath104 under @xmath105 where @xmath106 is monotonic and thus a homeomorphism from @xmath107 $ ] to @xmath108 $ ] .
furthermore if @xmath109 $ ] and @xmath51 is @xmath110 generic relative to @xmath111 on @xmath112 then @xmath113 is @xmath114 generic relative to @xmath102 on @xmath100 .
thus if this construction succeeds every @xmath115 $ ] is @xmath116 generic on @xmath117 $ ] for every @xmath118 as it s the image of some @xmath119 under @xmath120 which we abbreviate as @xmath121 . to ensure that @xmath122 is homeomorphic with @xmath82 harrington also required that @xmath123 and that @xmath124 .
now define @xmath125 to be @xmath126 and note that @xmath127 .
hence @xmath125 is clearly a continuous bijection between @xmath99 and @xmath128 .
the only remaining problem is to construct such a sequence .
the trick here is to observe that @xmath129 depends ( more or less ) only on @xmath130 and @xmath130 depends only on @xmath131 and so on .
thus @xmath122 can be built by looking only at @xmath132 .
this argument is nt too difficult to formalize in terms of the recursion theorem but unfortunately many of the important applications , including the ones we use later in this paper , depend on proving the result with an arbitrary computable ordinal @xmath2 substituted for @xmath91 so we must give the fully general construction . as we will see that while conceptually identical the technical details harrington avoided spelling out are definitely not trivial
[ thm : harrington - mcl - ordinal ] for every ordinal notation @xmath2 and tree @xmath133 there is a computable tree @xmath122 such @xmath117 $ ] and @xmath22 $ ] are homeomorphic and every @xmath115 $ ] is @xmath134 generic on @xmath22 $ ] . while this tells us how to copy @xmath135 down to lower trees it s no longer obvious how much we should copy . to extend harrington s argument to a sequence of length @xmath2
we will need to somehow specify an integer @xmath136 for every @xmath137 telling us how much of @xmath138 we should copy down to @xmath139 .
we need to ensure that @xmath140 will be the limit of @xmath141 for @xmath142 so @xmath117=[t_\alpha ] $ ] and that we can define @xmath143 from only @xmath144 to ensure our fixed point is non - empty so we must have : @xmath145\left ( \kleenelim{\lambda}{n } \kleeneleq \beta \kleenel \lambda \implies { \mathbrush{l}^{}(\kleenelim{\lambda}{n } ) } \leq { \mathbrush{l}^{}(\beta ) } \right)\label{edef : copylen - between } \\ & \lim_{n \to \infty } { \mathbrush{l}^{}(\kleenelim{\lambda}{n } ) } = \infty \\
\shortintertext{for concreteness we will also insist that } & \label{edef : copylen } { \mathbrush{l}^{}(\beta ) } = \begin{cases } 0 & \text{if } { { \beta}^{\diamond}}=\diverge\\ { \mathbrush{l}^{}({{\beta}^{\diamond}})}+n & \text{if } \beta=\kleenelim{{{\beta}^{\diamond}}}{n } \end{cases}\end{aligned}\ ] ] the difficulty in achieving these conditions is that in general a notation @xmath62 could appear at arbitrary places in the effective limit for arbitrarily many @xmath146 . a further difficulty is posed by the need to build a single function @xmath136 defined on a path through @xmath31 as required by some of the corollaries .
our strategy is to associate to each @xmath147 a unique limit notation @xmath148 to whose effective limit @xmath62 belongs .
since this proof is fairly technical we delay it s presentation until the appendix and blithely continue assuming we have a computable function @xmath136 satisfying the above ( below @xmath2 ) and that every @xmath62 appears in at most one effective limit ( below @xmath2 ) denoted @xmath148 ( extended to be total , increasing , limit valued ) .
this is slightly inaccurate , but we reserve those qualifications for the appendix .
this resolves the problem of how much to copy but we do nt yet know exactly what to copy . in the sketch of theorem [ thm : harrington - mcl ] we had a computable tree @xmath99 but in general at limit stages @xmath140 will only be computable in @xmath149 but we will still need to copy @xmath150 down to @xmath151 .
we show that we can always convert our trees to a form in which the segments requiring copying can always be uniformly recovered from the appropriate degree .
[ def : uniform - tree ] say the tree @xmath152 is uniform if @xmath153 is uniformly computable from @xmath154 for all @xmath26 .
converting @xmath155 into a uniform tree simply requires we delay killing branches in @xmath155 until they are long enough that we can use enough of @xmath149 to verify the branch gets killed .
[ lem : convert - to - n - bit ] suppose @xmath156 , @xmath157 and @xmath158 is uniform then there is a uniform @xmath159 with @xmath160=[\hat{t_\lambda } ] $ ] and @xmath161 .
furthermore , this holds with all possible uniformity .
as @xmath158 is uniform we may let @xmath162 without difficulty . now given @xmath163 with @xmath164 place @xmath165 unless some computation showing that @xmath166 converges in at most @xmath167 many steps while consulting only those columns of @xmath149 that encode @xmath168 for @xmath169 .
the uniformity is evident in the proof . for the remainder of the paper we will apply the preceding lemma without comment and assume without comment that any needed conversion of this kind
is done behind the scenes . in this section
we fix some ordinal notation @xmath2 and work to build a tower of trees @xmath170 as we sketched for @xmath91 ( technically speaking @xmath2 is nt truly arbitrary only the ordinal it denotes is ) .
we first describe the properties of the trees we seek to build . at each @xmath62
we will try to build our tree @xmath152 so that every path meets every @xmath171 $ ] set as soon as possible .
we capture the effect of this construction below with the notion of eagerly generic ( meaning @xmath97 generic over @xmath55 ) .
[ def : eager ] say that @xmath172 is eagerly generic ( over @xmath55 ) if for all @xmath80 $ ] and @xmath173{1 } $ ] there is some @xmath86 such that either @xmath19 super forces @xmath174 ( @xmath19 witnesses the @xmath175{1 } $ ] fact ) or @xmath19 super forces @xmath176 ( no @xmath71 ( not @xmath23 ) super forces @xmath174 ) .
note that this immediately entails that every @xmath177 $ ] is @xmath97 generic relative to @xmath55 on @xmath141 and the relation @xmath178 \psi $ ] is computable in @xmath179 for @xmath173{1 } \union \cpin[\zeron{\beta}]{1 } $ ] . since forcing on
@xmath141 always occurs relative to @xmath55 we will abbreviate @xmath180 $ ] as @xmath181 $ ] or even @xmath182 $ ] where this wo nt generate confusion .
we are now ready to precisely state the conditions our trees aim to satisfy .
[ def : downwardly - generic ] say that a sequence @xmath183 is a downwardly generic tower ( of length @xmath2 ) if for @xmath184 , @xmath152[def : downwardly - generic : beta - computable ] . for @xmath184 , @xmath141 is eagerly generic over @xmath55[def : downwardly - generic : eagerly ] .
if @xmath185 there is a monotonic embedding @xmath186 of @xmath138 into @xmath141 that induces a homeomorphism of @xmath187 $ ] and @xmath188 $ ] .[def : downwardly - generic : map ] [ def : downwardly - generic : map ] and [ def : downwardly - generic : beta - computable ] hold uniformly in @xmath62.[def : downwardly - generic : unifom ] @xmath189.[def : downwardly - generic : copying ] if @xmath34 a limit @xmath190[def : downwardly - generic : lim - unions ] note that by these last two conditions entail that @xmath157 . for
the remainder of this subsection we fix a downwardly generic tower @xmath183 of length @xmath2 and proceed to demonstrate it has the desired properties .
first we observe that our tower preserves the topological structure of @xmath191 and provides a moderately effective means of translation between trees at different levels .
[ lem : tower - homeomorphism ] for every @xmath192 there is a monotonic function @xmath193 uniformly computable in @xmath55 embedding @xmath141 into @xmath194 so as to induce a homeomorphism of @xmath188 $ ] and @xmath195 $ ] uniquely defined by the constraints : if @xmath196 then @xmath197 @xmath198[lem : tower - homeomorphism : extend - theta ] + if @xmath199 then @xmath200 [ lem : tower - homeomorphism : limits ] .
we use the method of effective transfinite recursion by assuming we have some index @xmath201 such that @xmath202 for every @xmath203 and build a computable function @xmath204 so that @xmath205 for every @xmath206 and then use the fixed point lemma to build a single computable function working for all @xmath207 .
the behavior of @xmath204 is spelled out plainly for @xmath208 a successor ( making use of part [ def : downwardly - generic : unifom ] to recover the various indexes ) and for @xmath208 a limit @xmath209 computes the the value @xmath210 where @xmath211 is the first such integer for which we observe @xmath212 . as @xmath213 the fixed point of @xmath204 plainly defines a monotonic function for all @xmath214 mapping @xmath194 into @xmath141 . by and
part [ def : downwardly - generic : copying ] of definition [ def : downwardly - generic ] any valid choice of @xmath167 yields the same result establishing uniqueness .
this leaves only the claim that @xmath215 gives a surjection of @xmath216 onto @xmath217 to verify .
assume not and let @xmath218 be the lexicographically least such that the claim fails for @xmath219 .
now clearly , by the minimality of @xmath220 and property [ def : downwardly - generic : map ] of definition [ def : downwardly - generic ] , @xmath221 evidently ca nt be a successor .
but if @xmath220 a limit , @xmath222 $ ] and @xmath223 then by the minimality of @xmath220 there is some some @xmath224 with @xmath225 . by condition [ lem : tower - homeomorphism : limits ] @xmath226 for all large enough @xmath167 so by monotonicity @xmath227 and by part [ def : downwardly - generic : copying ] of definition [ def : downwardly - generic ] @xmath228 .
thus @xmath229 is a path through @xmath194 with @xmath230 . with this lemma in mind
we adopt the notation that if @xmath52 is a path through @xmath122 then @xmath231 refers to the path in @xmath141 which maps to @xmath52 under @xmath232 which we abbreviate @xmath233 .
we now work toward showing @xmath122 will be sufficiently generic by translating @xmath234 , @xmath235 formulas about @xmath236 to @xmath175{1 } $ ] formulas about @xmath231 so that if @xmath231 forces the translated formula on @xmath139 then @xmath52 forces the original on @xmath122 .
our strategy will be to eliminate quantifiers from the interior of a formula by replacing @xmath237{1 } \union \cpin[\zeron{\gamma}]{1 } $ ] with the @xmath175{1 } $ ] formula asserting that @xmath238 forces @xmath174 when @xmath239 .
[ def : psi - star ] given a computable infinitary formula @xmath59 we define @xmath240 inductively as follows : @xmath241 \phi^{\beta}\right ) & \text{if } \phi^\beta \in \csigman[\zeron{\beta}]{1 } \cup \cpin[\zeron{\beta}]{1 } \\
\llor_{i \in \omega } \psi^{\beta+1}_{q(i ) } & \text { otherwise when } \phi=\llor_{i \in \omega } \psi_{q(i ) } \\
\lnot \psi^{\beta+1 } & \text { otherwise when } \phi=\lnot \psi \\ \end{cases } \\ \shortintertext{for \
( \lambda \ ) a limit } \phi^{\lambda } & = \begin{cases } \exists({\sigma \subset g_\lambda})\left({\vartheta^{\lambda}_{\gamma}}(\sigma ) \frc[{\gamma } ] \phi^\gamma\right ) & \text{if } \phi \in \csigman{\gamma } \union \cpin{\gamma } \text { for } \gamma \kleenel \lambda \\
\llor_{i \in \omega } \psi^{\lambda}_{q(i ) } & \text{otherwise when } \phi=\llor_{i \in \omega } \psi_{q(i ) } \\
\lnot \psi^{\lambda } & \text{otherwise when } \phi=\lnot \psi \end{cases } \end{aligned}\ ] ] [ lem : psi - star - forcing ] if @xmath242 then @xmath243 \phi^{\beta } $ ] iff @xmath244 \phi $ ] .
moreover , if @xmath245 then @xmath246{1 } $ ] fix @xmath220 to be least ordinal for which the equivalence @xmath247 \phi^{\gamma } \iff { \vartheta^{\gamma}_{}}(\sigma ) \frc[0 ] \phi $ ] fails to hold for some @xmath59 and let @xmath59 be the witness to this failure of least complexity .
so first suppose @xmath249 .
first suppose @xmath250{1 } \cup \cpin[\zeron{\beta}]{1 } $ ] . in this case
@xmath251 is defined as in the first case above so if @xmath252 \phi^{\beta+1 } $ ] then there must actually be some @xmath253 with @xmath254 \phi^{\beta } $ ] . by monotonicity
@xmath255 \phi^{\beta } $ ] and by the inductive hypothesis @xmath256 \phi $ ] .
alternatively suppose that @xmath257{1 } \cup \cpin[\zeron{\beta}]{1 } $ ] .
but if @xmath252 \llor_{i \in \omega } \psi^{\beta+1}_{q(i ) } $ ] then for some @xmath7 we must have @xmath252 \psi^{\beta+1}_{q(i ) } $ ] so by the minimality of @xmath59 we have @xmath258 \psi^{\beta+1}_{q(i ) } $ ] and therefore @xmath259 \phi^{\beta+1 } $ ] .
finally if @xmath260 and @xmath252 \lnot \psi^{\beta+1 } $ ] then again by the minimality of @xmath59 no @xmath261 in @xmath262 forces @xmath174 so therefore @xmath263 \lnot \psi=\phi $ ] .
consequently @xmath264 ca nt fail at @xmath249 .
going the other way if we suppose that @xmath250{1 } \cup \cpin[\zeron{\beta}]{1 } $ ] and that @xmath265 \phi $ ] then minimality ensures that @xmath255 \phi^{\beta } $ ] so by definition @xmath266 holds entailing @xmath252 \phi^{\beta+1 } $ ] . alternatively , suppose that @xmath267 \phi= \llor_{i \in \omega } \psi_{q(i ) } $ ] then for some @xmath7 we have @xmath267 \psi_{q(i ) } $ ] so by minimality of @xmath59 we infer @xmath252 \psi^{\beta+1}_{q(i ) } $ ] and therefore @xmath252 \phi^{\beta+1 } $ ] .
lastly if @xmath258 \phi= \lnot \psi $ ] then by other direction no extensions of @xmath19 on @xmath268 can force @xmath269 or some extension of @xmath270 would force @xmath174 so @xmath19 must force @xmath271 .
the proof for limit stages follows by the same considerations and the last claim follows by straightforward induction . [ lem : totally - generic ] every @xmath272 $ ] is @xmath134-generic on @xmath122 .
fix @xmath273 . by lemma [ lem :
psi - star - forcing ] @xmath274{1 } $ ] so either @xmath275 forces @xmath276 or @xmath277 as every path through @xmath278 is eagerly generic so applying lemma [ lem : psi - star - forcing ] again we conclude that @xmath279 forces either @xmath59 or it s negation on @xmath122 .
we now demonstrate the existence of a downwardly generic tower of length @xmath2
. our construction will begin with an arbitrary tree @xmath280 computable in @xmath281 which we will modify to be an eagerly generic tree @xmath191 .
from @xmath191 we will work downward to define @xmath141 for @xmath137 by way of the following effective process .
[ lem : tree - minus - one ] given trees @xmath282 and @xmath283 with @xmath284 there is a tree @xmath152 with and a monotonic function @xmath285 such that @xmath286.[lem : tree - minus - one : copy ] if @xmath287 then @xmath285 is a monotonic embedding of @xmath138 into @xmath141 such that @xmath288= [ t_\beta ] $ ] [ lem : tree - minus - one : map ] @xmath141 is eagerly generic.[lem : tree - minus - one : eagerly ] furthermore indexes for @xmath289 are computable from the indexes for @xmath138 and @xmath194 via a function that is total even when passed indexes for @xmath138 and @xmath194 that fail to converge on some values .
for simplicity we refer to @xmath138 as @xmath290 , @xmath285 as @xmath29 and @xmath141 as @xmath21 .
we fix a @xmath55 stagewise approximation to @xmath290 valid in the limit and ( implicitly using lemma [ lem : convert - to - n - bit ] ) we set @xmath291 .
we set @xmath29 to be the identity on @xmath292 and pause the entire construction at any stage where the approximation to @xmath290 does nt agree with @xmath292 .
thus should the condition @xmath293 fail , our construction eventually shuts down and refuses to produce a useful result .
thus we ve directly satisfied part [ lem : tree - minus - one : copy ] of the lemma .
for @xmath19 extending some element in @xmath292 we define @xmath294 to be the limit as @xmath295 goes to infinity of @xmath296 .
to ensure @xmath21 is computable from @xmath55 we decide whether @xmath19 is in @xmath21 at the first stage @xmath295 greater than the code of @xmath19 by placing it in @xmath21 if it is in the range of @xmath297 . at all times
we maintain that if @xmath296 is defined and @xmath298 then @xmath299 by letting @xmath296 become undefined if when required .
if at stage @xmath295 we observe some @xmath300 $ ] with code at most @xmath295 to be in @xmath301 and @xmath302 is defined but @xmath296 undefined we then also define @xmath303 $ ] .
we guarantee that [ lem : tree - minus - one : eagerly ] of the lemma holds by ensuring that if @xmath304 then either @xmath294 meets @xmath305 or for all @xmath306 with @xmath307 @xmath308 does not meet @xmath305 .
this is accomplished simply by letting @xmath309 be redefined to equal any @xmath307 with @xmath310{i } \isect t_s $ ] and reseting all @xmath311 for @xmath312 to undefined .
thinking of this as a finite injury argument we note that if @xmath313 eventually we reach some stage @xmath314 so that at any later stage @xmath295 , @xmath19 and all of it s initial segments are members of @xmath301 . furthermore if @xmath315 then we redefine @xmath296 no more than @xmath316 times after stage @xmath314 in attempts to meet in @xmath55 sets .
it is therefore clear that eventually @xmath296 will stabilize .
moreover , for the set of extensions of @xmath19 in @xmath21 to be infinite there must be infinitely many stages @xmath295 in which @xmath19 was in the range of @xmath297 so if @xmath317 $ ] then @xmath19 is in the range of @xmath29 thus part [ lem : tree - minus - one : map ] of the lemma is satisfied .
the uniformity is evident from the proof , but some remarks about why the resulting function is total even when passed partial indexes is warranted .
with respect to @xmath138 all that is really necessary is that eventually all members of @xmath138 stay in the approximation while non - members are out of the approximation at infinitely many stages so being in @xmath179 would suffice . since lemma [ lem : convert - to - n - bit ] only cared about elements being enumerated into the compliment of @xmath318 that index may also be partial .
note that the condition @xmath287 in the above lemma is guaranteed to be satisfied if @xmath138 properly copies @xmath319 by lemma [ lem : build - copylen ] .
also remember that should @xmath320 we defined @xmath321 to be another copy of @xmath278 . since we only make use of @xmath321 to copy @xmath322 we may safely pretend ( by redefinition ) that @xmath323 whenever it would otherwise be larger than @xmath2 .
[ lem : downward - tower - exists ] given @xmath133 there is a downwardly generic tower @xmath183 of length @xmath2 with @xmath324 the effectively given image of @xmath21 . by the same argument given in lemma [ lem : tree - minus - one ] we can effectively transform @xmath21 into an eagerly generic @xmath325 .
furthermore we may assume that @xmath2 is a limit ordinal during construction by applying lemma [ lem : tree - minus - one ] finitely many times until we reached a limit level .
now fix an index @xmath7 for @xmath278 and let @xmath326 be the computable function giving an index for @xmath141 given an index @xmath327 for @xmath138 and @xmath328 for @xmath321 .
we now define a computable function @xmath204 to behave as follows with the intent that @xmath329 should define a function from notations @xmath330 to an index for @xmath141 whenever @xmath331 defines the same function on @xmath332 .
@xmath333 fix @xmath201 to be a fixed point of @xmath204 and let @xmath141 be the tree defined by index @xmath334 relative to @xmath55 for @xmath137 . note that it is enough to show that @xmath141 is built as per lemma [ lem : tree - minus - one ] from @xmath138 and @xmath321 since lemma [ lem : build - copylen ] ensures that if @xmath335 then @xmath336 and as well as that @xmath155 for @xmath34 a limit is the limit of @xmath141 with @xmath142 [ def : downwardly - generic ] .
now suppose that @xmath141 fails to be defined or satisfy the conditions of lemma [ lem : tree - minus - one ] with respect to @xmath138 and @xmath321 . since there are no infinite decreasing sequences of ordinals we can assume that @xmath337 where @xmath338 is defined and satisfies the conclusions of lemma [ lem : tree - minus - one ] .
thus if @xmath339 by the choice of @xmath201 as a fixed point we also have @xmath340 so @xmath141 is defined by the application of lemma [ lem : tree - minus - one ] .
note that the work here is really being done by lemma [ lem : tower - homeomorphism ] which verified that merely being the image of a monotonic function and the properties of the function @xmath136 ensure that all trees in the tower are homeomorphic .
this completes the proof of theorem [ thm : harrington - mcl - ordinal ] . at this point
it is interesting to note that this is in some sense optimal since every member of a countable hyperarithmetic class @xmath341 is itself a hyperarithmetic singleton .
[ def : alpha - root ] say @xmath21 is the @xmath2-reduct of @xmath290 if @xmath101 where @xmath122 is constructed as described above from @xmath342 .
also we call those @xmath272 $ ] an @xmath2 root of @xmath343 $ ] if @xmath52 is the image of @xmath344 under the constructed homomorphism
. note that an index for the @xmath2-reduct of @xmath290 as a computable set can be effectively computed from a index for @xmath345 as a @xmath346 computable set .
harrington observed several other important consequences of the above method in @xcite that have also never been formally published and we take the time to present those that can be stated in terms of classical computability theory here and leave those about admissible sets and various implications in second order arithmetic to another paper . [ def : subgeneric ] following harrington @xcite we say a degree @xmath347 is @xmath2 subgeneric for @xmath348 if for all @xmath137 @xmath349 satisfies both @xmath350[def : subgeneric : join - jump ] @xmath351\left ( x \tleq \zeron{\alpha } \land x \tleq \jumpn{\tdeg{d}}{\beta } \implies x \tleq \zeron{\beta } \right ) $ ] [ def : subgeneric : meet - zeron - beta ] a version of harrington s first corollary in @xcite can now be stated .
[ cor : subgeneric ] for each @xmath352 there is a sequence @xmath353 so that for all @xmath354 @xmath355 is @xmath2 subgeneric .
@xmath356[cor : subgeneric : alpha - independence ] .
@xmath357 is a the unique solution of a @xmath0 formula the index for which is given uniformly in @xmath26 .
our first task is to assure ourselves we already know how to satisfy part [ def : subgeneric : join - jump ] of definition [ def : subgeneric ] .
[ lem : alpha - subgeneric - part-1 ] if @xmath52 is an @xmath2 root then @xmath52 satisfies part [ def : subgeneric : join - jump ] of the .
fix @xmath183 witnessing that @xmath52 is an @xmath2 root and @xmath358 and @xmath359 such that @xmath360 . by lemma [ lem :
psi - star - forcing ] if @xmath272 $ ] then @xmath361 \psi^\lambda $ ] . since either @xmath362 or it s negation
is super forced on @xmath363 by @xmath364 we can compute @xmath365 from @xmath366 completing the proof . building @xmath52 as an @xmath2 root that also satisfies part [ def : subgeneric : meet - zeron - beta ] of the requires slightly more work . given @xmath367 and @xmath368 these computations
must be super forced on @xmath138 to be equal but we need to guarantee they are super forced to agree on @xmath141 to ensure @xmath369 . since @xmath141 is nt the image of @xmath138 under @xmath285 super forcing on @xmath138 does nt translate to super forcing on @xmath141 so we must guarantee this occurs manually . since @xmath141 lacks access to @xmath179 we ca nt directly diagonalize but must instead try to preserve disagreeing options for the computation of @xmath42 from @xmath370 and let the diagonalization occur on @xmath138 .
we first must ensure that @xmath138 leaves options open that can be extended on @xmath141 to incompatible computations .
[ def : padded - tree ] say @xmath141 is padded if whenever @xmath371 , @xmath372 then @xmath373 , \sigma\concat[1 ] \in t_\beta $ ] . and now give conditions that ensure these incompatible computations exist .
[ def : disagreement - eager ] say that @xmath374 is disagreement preserving if whenever @xmath373 , \sigma\concat[1 ] \in t_{\beta+1 } $ ] and @xmath375 then either @xmath376 ( \recfnl{i}{\zeron{\beta}}{\tau } \compat \recfnl{i}{\zeron{\beta}}{\tau ' } ) \\ \shortintertext{or } & \recfnl{i}{\zeron{\beta}}{\theta^{\beta+1}(\sigma\concat[0 ] ) } \incompat \recfnl{i}{\zeron{\beta}}{\theta^{\beta+1}(\sigma\concat[1 ] ) } \end{aligned}\ ] ] [ def : subgeneric - root ] say that @xmath183 is a disagreement preserving downwardly generic tower if it is a downwardly generic tower and for each @xmath62 with @xmath377 , @xmath139 is padded and @xmath285 is disagreement preserving .
we define the notions of an @xmath2 subgeneric - reduct and @xmath2 subgeneric - root by modifying definition [ def : alpha - root ] to use disagreement preserving downwardly generic towers .
note that we can produce @xmath2 subgeneric - reducts with the same degree of effectivity as we enjoyed for @xmath2 reducts .
[ lem : subgeneric - root ] if @xmath52 is a @xmath2 subgeneric - root then @xmath52 is @xmath2 subgeneric . by lemma [ lem : alpha - subgeneric - part-1 ]
it is sufficient to show that @xmath52 satisfies part [ def : subgeneric : meet - zeron - beta ] of definition [ def : subgeneric ] .
suppose , for a contradiction , that @xmath52 fails this condition for the set @xmath42 and let @xmath184 be the least such that @xmath369 and for some @xmath378 we have @xmath379 but @xmath380 .
assume @xmath62 is a successor then we must have @xmath381 or @xmath62 would not have been the least failure .
now fix @xmath382 such that @xmath383 let @xmath174 be the @xmath384{1 } $ ] formula asserting that these computations disagree .
since the computations agree we have @xmath385 so pick @xmath386 such that @xmath387 super forces @xmath176 on @xmath388 where @xmath389 \neq \recfnl{j}{\zeron{\gamma } \tplus \theta^{\beta}(\sigma ) } { x}\conv[s]\right)\ ] ] we now work to define an initial segment @xmath390 of @xmath391 extending @xmath19 so that @xmath318 will preserve any potential disagreement so it s observed between the inputs @xmath392 ) $ ] and @xmath393 ) ] $ ] if ever . for this purpose pick @xmath394 so that @xmath395 and set @xmath396 . now
if @xmath397 ) } \incompat \recfnl{i}{\zeron{\gamma}}{\theta^{\gamma+1}(\sigma'\concat[1 ] ) } $ ] then for some choice of @xmath398 the string @xmath399 ) $ ] disagrees with @xmath42 so @xmath400 \models \psi $ ] contradicting the assumption that @xmath19 super forced @xmath401 on @xmath388 .
thus by definition [ def : disagreement - eager ] every extension of @xmath402 yields compatible computations under @xmath403 . thus given @xmath404
we may compute @xmath405 as the value of the first converging computation @xmath406 for some @xmath407 with @xmath408 .
since @xmath409 this search can be performed computably in @xmath168 and as @xmath410 eventually a converging computation will always be found .
now suppose @xmath62 is a limit and again let @xmath378 and @xmath411 satisfy . by the minimality of @xmath62 ,
if @xmath412 with @xmath413 we are done so without loss of generality we may assume that @xmath414 . to coordinate the behavior of the computation of @xmath42 from the various @xmath415
we show that we can choose a single index for all such computations . by lemma [ lem : tower - homeomorphism ]
we can uniformly compute @xmath416 from @xmath417 using @xmath418 .
we claim that there is a single index @xmath328 such that @xmath419 whenever @xmath420 .
the computation coded by @xmath328 can check whether @xmath421 by inspecting @xmath422 allowing @xmath423 to recover @xmath424 at which point it can apply @xmath425 . note that our index @xmath328 has the property that @xmath426 provided @xmath427 and @xmath428 .
armed with this index we define @xmath59 asserting that some such computation for @xmath42 disagrees with the computation from @xmath55 .
@xmath429 \neq \recfnl{j'}{\zeron{\kappa } \tplus { \vartheta^{\beta}_{\kappa}}(\sigma ) } { x}\conv[s]\right)\ ] ] since @xmath59 is false let @xmath430 on @xmath139 force @xmath67 .
now fix @xmath431 so that @xmath432 and let @xmath433 . now
if @xmath434 by @xmath435 and by @xmath436 for @xmath437 . now if @xmath438 then we also have @xmath439 and @xmath400 \in t_{\kappa+1}\restr{{\mathbrush{l}^{}(\kappa+1)}}=t_{\beta}\restr{{\mathbrush{l}^{}(\kappa+1 ) } } $ ] . turning our attention back to definition [ def : disagreement - eager ]
suppose that @xmath440 ) } $ ] and @xmath441 ) } $ ] are incompatible then we can choose @xmath442 $ ] such that @xmath443 is incompatible with @xmath42 .
this holds as @xmath444 can be factored to @xmath445 and as @xmath446 simplifies to just @xmath447 .
as such a @xmath308 would force @xmath59 we conclude that every @xmath448 extending @xmath449 yields compatible computations and we compute @xmath42 from @xmath422 as we did in the successor stages . hence @xmath412 contradicting the minimality of @xmath62 .
we now show that a slight modification of the construction of @xmath139 from @xmath138 and @xmath321 we performed above lets us build a subgeneric @xmath2 root .
[ lem : tree - minus - one - alt1 ] the following conditions may be added to those of lemma [ lem : tree - minus - one ] so that @xmath139 continues to be effectively built from @xmath138 and @xmath321 while jointly satisfying all conditions . if @xmath450 and @xmath194 is padded or @xmath451 then @xmath141 is padded .
@xmath285 is disagreement preserving[lem : tree - minus - one : disagreement - preserving ] to ensure that part [ lem : tree - minus - one : disagreement - preserving ] holds whenever @xmath375 , @xmath452 ) $ ] and @xmath453 ) $ ] are both defined and @xmath454{i}{\tau_0 } { } \compat \recfnl[s]{i}{\tau_1 } { } $ ] , but there are @xmath455 with @xmath454{i}{\tau'_0 } { } \incompat \recfnl[s]{i}{\tau'_1 } { } $ ] then set @xmath456)=\tau'_0 $ ] and @xmath457)=\tau'_1 $ ] and unset @xmath458 for every @xmath459 . to ensure that @xmath141 is padded we simply place @xmath373 $ ] and @xmath460 $ ] into @xmath141 whenever @xmath371 and for some @xmath167 @xmath461 . if @xmath451 and this suffices .
otherwise @xmath141 is padded as @xmath462 and @xmath194 is padded . note that given an initial tree @xmath21 we can easily perform effective modifications to ensure it is padded so substituting lemma [ lem : tree - minus - one - alt1 ] into the construction given by lemma [ lem : downward - tower - exists ] yields a disagreement preserving downwardly generic tower of length @xmath2 . thus given a @xmath281 index for @xmath463 viewed as a tree @xmath191 with @xmath464=\set{g_\alpha } { } $ ] we can compute the index for a computable tree @xmath122 with a unique path @xmath52 which by lemma [ lem : subgeneric - root ] is @xmath2 subgeneric . while this easily gives a ( uniformly witnessed ) sequence of @xmath0 singletons @xmath465 of @xmath2 subgenerics
this is not quite sufficient to prove corollary [ cor : subgeneric ] as we must still ensure that part [ cor : subgeneric : alpha - independence ] of definition [ cor : subgeneric ] holds . to do this
we observe [ lem : uniform - alpha - independence ] there is a uniform sequence @xmath466 each having a unique path @xmath467 such that @xmath468 our construction builds @xmath469 as the limit of @xmath470 via a finite injury argument .
the requirements @xmath471 demand that @xmath472 and are met by changing the value of @xmath473 to disagree with the computation in question whenever such a change is not restrained by a higher priority requirement and restraining any changes in the use of this computation or of @xmath474 .
every time @xmath473 is set to a new value it is picked large enough not yet to have been enumerated into the compliment of @xmath475 . this now suffices to complete the proof of corollary [ cor : subgeneric ] . using the uniformity of the trees @xmath476 from lemma [ lem : uniform - alpha - independence ] and the uniformity of the construction in lemma [ lem : downward - tower - exists ] ( using the modified lemma [ lem : tree - minus - one - alt1 ] ) we get a uniform sequence of computable trees @xmath477 each containing a single @xmath2 subgeneric path @xmath465 . to see that part [ cor : subgeneric : alpha - independence ] of definition [ cor : subgeneric ] holds observe that the uniformity of the trees @xmath476 and the uniform definition of the maps @xmath478 guarantees the equivalence @xmath479 holds uniformly .
also by the uniformity of lemma [ lem : alpha - subgeneric - part-1 ] @xmath480 holds uniformly so if part [ cor : subgeneric : alpha - independence ] failed we would have the contradiction @xmath481 it is worth remarking that the result claimed by harrington in @xcite is nt actually lemma [ cor : subgeneric ] but the substantially stronger version below .
[ cor : subgeneric - star ] for each @xmath352 there is a sequence @xmath353 so that for all @xmath354 1 .
@xmath355 is @xmath2 subgeneric .
@xmath482[cor : subgeneric - star : alpha - independence ] .
@xmath357 is a the unique solution of a @xmath0 formula the index for which is given uniformly in @xmath26 .
corollary [ cor : subgeneric - star ] replaces which required that no @xmath357 could be computable in the join of the @xmath2 jumps of the remaining @xmath465 with the substantially stronger requirement that @xmath357 not be computable in the @xmath2 jump of the join of the remaining @xmath465 . corollary [ cor : subgeneric - star ] is true but we have only been able to prove the result by making some substantial modifications to the underlying framework which we sketch below . to establish we introduce a notion of mutual genericity for a sequence of the reals @xmath483 on the sequence of trees @xmath484 where @xmath485 is a path through @xmath486 . in particular we define @xmath487 to be the function where @xmath488 and write @xmath489 for @xmath490 .
we further define @xmath491 to be the tree consisting of those nodes @xmath492 with @xmath493 for every @xmath26 and say that the sequence of singletons @xmath494 is mutually @xmath2 generic on @xmath484 if @xmath495 is @xmath2 generic on @xmath496 . we prove [ cor : subgeneric - star ] by simultaneously building @xmath91 many disagreement preserving downwardly generic towers consisting of the trees @xmath497 for @xmath184 where @xmath498 is the unique path through @xmath498 and the trees @xmath497
satisfy the obvious generalization of being eagerly generic to the notion of mutually eagerly generic .
we stipulate our coding function has the property that @xmath499 is always greater than @xmath500 so if @xmath501 we may assume that @xmath502 with each @xmath503 and all but finitely many of them equal to the empty string generalizing our previous construction we now build @xmath497 as the image of @xmath504 mapping @xmath505 to @xmath497 .
naively one might imagine that we could straightforwardly carry out the same forcing construction we used previously but now applied to @xmath506 as in the standard ( not localized to a tree ) product forcing construction .
however , since we wish to maintain @xmath507 so as to still produce a sequence @xmath485 of @xmath2-subgeneric roots such a simple argument wo nt suffice .
in particular by demanding that every path through @xmath506 extending @xmath492 also pass through @xmath508 we would impose pruning on the factors @xmath497 which would in turn force a pruning of @xmath506 above other nodes @xmath509 even when @xmath510 is incompatible with @xmath492 because we could still have @xmath511 . to avoid this difficulty we ensure that the paths in @xmath497 carry with them the information about the paths in @xmath512 .
in particular we will ensure that if @xmath513 then @xmath514 for every @xmath7 . as before we define @xmath515 as the limit of a stagewise construction so we suppose that @xmath516 is defined and @xmath517 and define @xmath518 for each @xmath7 exactly as we did in lemma [ lem : tree - minus - one ] . note that if @xmath519 this has the effect of duplicating the usual action of @xmath520 as it would apply to @xmath521 in the single tree @xmath522 at @xmath523 on @xmath524 for all @xmath525 .
it is readily apparent that this definition ensures the required incompatibility property mentioned above .
we may now safely prune branches on @xmath506 without fear of inadvertent interference .
we now simply demand that if at some stage @xmath526 we discover some @xmath527 , @xmath528 with some finite ( contiguous ) initial segment of @xmath529 meeting @xmath530 then we redefine @xmath518 to be equal to @xmath531 and unset @xmath531 for every @xmath532 . by our incompatibility property
this ca nt have any effect on any node in @xmath506 incompatible with @xmath529 . at this point
we may now appeal to the fact that @xmath533 is the unique path through @xmath505 and argue that @xmath534 extends to an infinite path on @xmath497 if and only if @xmath521 extends to an infinite path on @xmath522 for all @xmath26 .
a similar approach can be applied to demand a slightly modified version of disagreement preservation . while this construction does nt yield a @xmath179 computable homeomorphism from @xmath535 $ ] to @xmath536 $ ] ( since some nodes would need to be mapped to many potential values ) it does yield a @xmath179 homeomorphism from @xmath537 $ ] to @xmath538 $ ] and this is sufficient to give both the desired mutual genericity as well as preserve the desired properties of the non - mutual construction .
this completes our sketch of corollary [ cor : subgeneric - star ] . before we finish our discussion of harrington s work in @xcite one final corollary is worth mentioning .
[ cor : totally - subgeneric ] there is a non - empty @xmath0 class whose members are @xmath2 subgeneric for every @xmath352 . in the appendix
we establish the existence of a @xmath539 linearly ordered set of notations @xmath540 cofinal in a path through @xmath31 such that the functions @xmath541 are uniformly computable in @xmath2 on the set @xmath542 for @xmath543 .
now consider the predicate @xmath544 consisting of those sets @xmath21 coding computable trees that with @xmath101 in some disagreement preserving downwardly generic tower of length @xmath2 with @xmath545 not having well - founded height less than the notation denoted by @xmath2 .
it is easily checked that @xmath544 is @xmath546 uniformly in @xmath2 as it is easily defined via number quantification over @xmath547 since the set of notations whose height is less than that of @xmath2 is uniformly computable from @xmath547 .
thus we may safely identify @xmath543 with the @xmath546 index for @xmath548 .
moreover @xmath549 entails @xmath550 and clearly @xmath551 .
now fix some @xmath546 subset @xmath552 of @xmath540 and consider @xmath553 . by
@xmath554 bounding there is some @xmath543 bounding @xmath552 so this intersection contains the non - empty collection @xmath548 .
thus by kreisel s compactness theorem @xcite there is some @xmath21 in @xmath555 and by lemma [ lem : subgeneric - root ] every path through @xmath21 is @xmath2 subgeneric for every @xmath352 .
moreover , @xmath21 does nt have well - founded height below @xmath556 so as @xmath21 is computable there must be some infinite path through @xmath21 . indeed , since no path through @xmath21 is hyperarithmetic @xmath22 $ ] must be a perfect set .
interest in the computational properties of fast growing functions goes back to post s program and the realization that how fast the enumeration of @xmath557 grows is a measure of the thinness of @xmath557 , and as rice first showed @xcite when he characterized the hyperimmune sets this way , it s often an easier concept to work with .
with seeming ingratitude this approach soon turned on post giving yates @xcite the tools he needed to put the nails in the coffin of post s program by building a complete maximal set .
later martin improved this analysis to fully characterize the degrees of maximal sets @xcite and even today studying the relation between rate of growth and computational power continues to pay off @xcite .
strangely , however , while many different notions of ` fast growing ' have been proposed and the degrees of such functions ( partially ) characterized little work has been done in the other direction . that is given a degree how fast much a function grow to compute that degree ? to this end we follow slaman and groszek in introducing the following definitions @xcite . [ def : moduli - of - computation ] the function @xmath558 is a moduli ( of computation ) for a degree @xmath349 if every @xmath559 computes @xmath349 .
@xmath560 is a uniform moduli of computation if there is some fixed computable functional @xmath561 and @xmath562 such that @xmath563 .
if furthermore @xmath560 is a ( uniform ) moduli of computation for @xmath564 we say that @xmath560 is a ( uniform ) self - moduli .
it s natural to respond to this definition by first asking when can a degree @xmath349 even have a moduli of computation ? what about a uniform moduli of computation
? can any degree be computed ( uniformly ? ) by sufficiently fast growing functions ?
though this side of the relationship between rates of growth and computational power hasnt received as much attention as it s opposite these questions are natural enough they have multiple published solutions that are disguised by terminological differences .
we first look to the uniform case where one can show the degrees with a uniform self - moduli are just the @xmath0 singletons ( in @xmath76 ) we direct the reader to jockusch and mclaughlin @xcite for the earliest easily straightforward english language proof but follow them in crediting kuznecov and trahtenbrot @xcite and latter myhill @xcite .
we generalize this result to those functions @xmath560 with a uniform moduli in some computable ordinal number of jumps .
informally the relationship is simply that @xmath560 is a @xmath565 singleton if and only if the natural fast growing function computable in @xmath566 is a uniform modulus for @xmath560 . to state the theorem formally we need to replace `` natural fast growing function '' with an explicit function .
[ def : fastn ] @xmath567\left(\recfnl{i}{\zeron{\beta}}{i}\conv \iff \recfnl{i}{\zeron{\beta}}{i}\conv[t ] \right ) } & \text{otherwise } \end{cases } \\ & { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}{\beta+1}(x)=\max(\hat{\xi}^{\beta + 1}(x ) , \sup_{\gamma \kleeneleq \beta } { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}{\gamma}(x ) ) \\ \shortintertext{for \
( \lambda \kleeneleq \alpha \ ) a limit } & { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}{\lambda}(x ) = \sup_{\beta \kleenel \lambda } { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}{\beta}(x ) = \sup_{\substack { \beta \kleenel \lambda \\ \gcode{\beta } \leq x } } { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}{\beta}(x)\end{aligned}\ ] ] we relativize this notion by setting @xmath568 and building @xmath569 as above . note that as the notations below some given notation @xmath62 can be effectively computed from @xmath570 these supremums can be easily deciphered by any set computing @xmath570 . with this in mind
the following properties should be straightforward to verify so are presented without proof .
[ lem : fastn ] for each @xmath184 , @xmath558 @xmath571 is uniformly computable from @xmath566 and has the following properties .
@xmath572 is a uniform self - modulus for @xmath55 and the functional witnessing this uniformity is itself uniform in @xmath62 .
@xmath573\left ( { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}(h){\gamma } { \ll}{\let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}(h){\beta } \right ) $ ] .
there is a stagewise approximation @xmath574(h){\beta+1 } $ ] uniformly computable in @xmath566 and strictly increasing in @xmath295 such that @xmath575{\beta+1}(x)={\let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}(h){\beta+1}(x ) $ ] .[lem : fastn : increasing ] the predicate @xmath576 is @xmath577.[lem : fastn : sigman - geq ] note that all of the above relativizes to @xmath578[h]{\beta } $ ] .
we can now formally describe the general relation between @xmath565 singletons and uniform moduli .
[ thm : uniform - modulus - in - beta - jumps ] @xmath571 is a uniform modulus for @xmath560 if and only if @xmath560 is a @xmath565 singleton .
suppose @xmath571 is a uniform modulus for @xmath560 witnessed by the reduction @xmath561 . by part [ lem : fastn : sigman - geq ] of lemma [ lem : fastn ] we note that there is a @xmath577 formula @xmath579 asserting that there is some @xmath580 and @xmath581 .
thus the formula @xmath84 defined below is equivalent to a @xmath582 formula @xmath583\forall[s]\forall[x]\left(\recfnl[s]{}{\sigma}{x}\diverge \lor \recfnl[s]{}{\sigma}{x}=g(x ) \lor \psi(\sigma ) \right)\ ] ] clearly @xmath584 holds as if @xmath585 and @xmath586 it must have value @xmath587 .
but if @xmath588 then pick @xmath589 with @xmath590 and @xmath591 where @xmath592 . such @xmath19 witnesses @xmath593
so @xmath560 is a @xmath565 singleton .
conversely suppose that @xmath594 has unique solution @xmath560 .
we may put @xmath59 in the form @xmath595\psi(\sigma ) $ ] where @xmath174 is a @xmath596 formula not mentioning @xmath560 .
now given @xmath597 we can uniformly recover @xmath55 from @xmath52 and thus compute a tree @xmath21 consisting of all those @xmath598 for which @xmath579 holds .
as @xmath599 @xmath560 is a path through @xmath21 and as any path through @xmath21 would satisfy @xmath59 it is unique . as @xmath21 is finitely branching we can avail ourselves of kenig s lemma to establish that @xmath52 uniformly computes @xmath560 . note that the specific form of @xmath600 is nt important only that @xmath601 and that @xmath600 is a uniform modulus for @xmath179 .
this suffices to give a uniform modulus for every hyperarithmetic function and it is easy to see ( as in @xcite ) that @xmath558 has a uniform modulus if and only if @xmath560 is @xmath546 .
while solovay finally classified those functions with some modulus of computation in @xcite using a different method we follow the approach taken in @xcite using hechler style forcing conditions to demonstrate the following lemma .
[ lem : mod - imp - uniform ] if @xmath52 has a modulus of computation than @xmath52 has a uniform modulus of computation we quickly sketch the proof .
the conditions will be cohen style conditions in @xmath11 paired with some @xmath602 we commit to majorizing .
if @xmath560 is a modulus for @xmath52 then the forcing conditions do their best to produce some @xmath559 not computing @xmath52 and their failure can only occur if there is some sufficiently fast growing @xmath603 above which the reduction is uniform . combining this result with the remark above yields
solovay s result .
[ thm : modulus - hyper ] @xmath558 has a modulus of computation if and only if @xmath560 is @xmath546 .
the above results pose a very puzzling question : all the natural examples of moduli are uniform moduli and every function with a modulus must have a uniform modulus so can the two notions come apart and if so by how much ?
the remainder of this paper is devoted to showing that these two notions come apart as far as possible .
in particular we prove the following result .
[ thm : nonu ] for each @xmath348 there is a a self - modulus @xmath604 such that no @xmath605 for any @xmath137 is a uniform modulus for @xmath606
. we can now prove [ thm : nonu ] .
we proceed by fixing some notation @xmath2 and describe in this section how to build a self - modulus @xmath606 with no uniform moduli computable from any @xmath607 .
this requires walking a careful line between making @xmath606 unique enough that every faster growing function computes it but not so unique that they can do so uniformly .
our approach is to build @xmath606 as a highly ` generic ' function that is nevertheless unique for all ` small ' functions majorizing it .
any ` large ' function majorizing @xmath606 will have enough computational power to watch our construction of @xmath606 while the uniqueness of @xmath606 relative to the ` small ' functions majorizing @xmath606 will let them compute @xmath606 .
essentially large will mean dominating @xmath608 and @xmath606 will be built computably in @xmath281 leaving the rest of the construction to deal with small functions and to be sufficiently generic to avoid small uniform moduli .
the level of genericity required is given by the next lemma .
[ lem : generic - not - unif - sm ] if @xmath52 is @xmath2 generic on @xmath609 and non - isolated then no @xmath610 is a uniform modulus for @xmath52 . for contradiction fix @xmath611 as in the lemma , @xmath612 a truth - table functional and @xmath613 a turing functional such that @xmath614 and for every @xmath559 @xmath615 .
now let @xmath59 be the @xmath234 formula defined below asserting that for some @xmath616 @xmath617 disagrees with @xmath52 .
@xmath618 \neq g(x)\right)\ ] ] if @xmath84 then there would be some @xmath619 extending @xmath308 so @xmath620 .
thus @xmath84 and as @xmath621 we must have some @xmath622 with @xmath60 \phi $ ] . as @xmath52 is non - isolated we can fix another path @xmath623 on @xmath21 extending @xmath19 .
let @xmath624 which as @xmath612 is a truth table reduction must be total . since @xmath625 we know @xmath626 holds .
now fix some @xmath627 . by assumption
@xmath628 contradicting the fact that @xmath629 agrees with @xmath344 everywhere both are defined .
thus , we can make @xmath606 sufficiently generic for our purposes by building it as a non - isolated path through some @xmath2 reduct @xmath122 .
if we had simply made @xmath606 fully @xmath2 generic then it would nt be a self - modulus at all since if @xmath52 is a non - isolated path through @xmath21 and @xmath630-generic on @xmath21 _ relative _ to @xmath21 then @xmath52 is not a self - modulus .
ideally we would simply manipulate @xmath122 so that if some @xmath631 $ ] then @xmath632 ensuring that if @xmath633 either @xmath634 has unique path @xmath606 or @xmath635 . however , @xmath122 must be computable so this condition is too strong .
instead we will impose a scrambled version of this condition .
[ def : uniquely - small ] a tree @xmath21 is uniquely @xmath62 small if @xmath636}]\forall[x]\left ( f(x ) , g(x ) \leq { \let\@pmg@parenarg\@pmg@undefined\let\@pmg@braketarg\@pmg@undefined{\xi^{}\ifdefined\@pmg@parenarg , \@pmg@parenarg\fi } \ifdefined\@pmg@braketarg _ { \@pmg@braketarg}\fi}{\beta}(x ) \implies f\restr{x+1}=g\restr{x+1 } \right)\ ] ] ultimately we must project the impact of making @xmath138 uniquely @xmath62 small down to @xmath122 without while retaining the ability to extract @xmath179 from @xmath55 and @xmath560 where @xmath637 for some @xmath638 $ ] .
this requires we further restrict our choice of reduction functions @xmath285 .
[ def : largness - preserving ] say a monotonic map @xmath639 is largeness preserving if every @xmath640 is non - decreasing and if @xmath641)=\sigma'\concat\tau ' $ ] then every @xmath642 satisfies @xmath643 .
[ def : uniquely - small - tower ] say a downwardly generic tower @xmath644 is a uniquely small tower ( of length @xmath2 ) if every @xmath141 is uniquely @xmath62 small and every @xmath285 in part [ def : downwardly - generic : map ] of definition [ def : downwardly - generic : map ] is largeness preserving .
say @xmath122 is an @xmath2 uniquely small reduct if it occurs in some uniquely small tower of length @xmath2 .
we now fix a uniquely small tower @xmath644 of length @xmath2 for the remainder of the proof .
[ lem : treemap - largeness - preserving ] for every @xmath184 the monotonic function @xmath232 is largeness preserving .
suppose that @xmath62 is the least failure . if @xmath645 then @xmath646 .
it is straightforward to verify that the composition of two largeness preserving functions is largeness preserving yielding the contradiction .
now suppose @xmath62 is a limit . since @xmath647 is largeness preserving we know that @xmath232 is non - decreasing . but
@xmath648 ) = { \vartheta^{\gamma}_{0}}(\sigma\concat[i ] ) $ ] for some @xmath649 by part [ lem : tower - homeomorphism : limits ] of lemma [ lem : tower - homeomorphism ] so by the minimality of @xmath62 @xmath648 ) $ ] ca nt fail the other half of definition [ def : largness - preserving ] .
while the motivation for making @xmath285 largeness preserving is to protect the encoding of @xmath281 in every pair of @xmath650 $ ] it also provides the following useful property .
[ lem : all - preimages - below ] a bound @xmath651 on @xmath652 for those @xmath19 with @xmath653 and @xmath654 can be uniformly computed from @xmath655 .
let @xmath656 be the set containing the empty string and let @xmath657 .
@xmath658}{\sigma \in s_n \land \exists({x \leq l(n)+1})\left(h(x)\geq i\right ) \land \sigma\concat[i ] \in t_{\beta+1 } }
\\ l(n+1 ) & = \max_{\sigma \in s_{n+1 } } \lh{{\vartheta^{\beta}_{}}(\sigma ) } \end{aligned}\ ] ] note that by definition [ def : largness - preserving ] in defining @xmath659weve only excluded values of @xmath7 that guarantee @xmath660 ) $ ] .
[ lem : singleton - or - jump ] suppose @xmath661 is a path through an @xmath2 uniquely small reduct @xmath122 then @xmath606 is a self - modulus
. let @xmath633 which without loss of generality we may assume is non - decreasing .
first suppose that @xmath606 is the only path through @xmath122 satisfying @xmath633 .
in this case let @xmath290 be the set of @xmath662 with @xmath663 .
clearly @xmath290 is a tree and @xmath606 is the unique path through @xmath664 . as @xmath290 is a finitely branching tree computable in @xmath560 , kenig s lemma lets us uniformly compute @xmath606 from @xmath560 .
so suppose @xmath665 with @xmath666 and @xmath667 .
we argue by effective transfinite recursion that @xmath668 ( uniformly in @xmath62 ) for @xmath669 leaving the routine details for the reader . at limit stages
the induction is straightforward so suppose @xmath668 .
we show that given @xmath670 @xmath560 can ( uniformly ) compute @xmath671 and thus @xmath672 . by way of lemmas
[ lem : treemap - largeness - preserving ] and [ lem : all - preimages - below ] we compute @xmath167 such that if @xmath673 and @xmath674 then @xmath675 .
we now verify @xmath671 .
since @xmath676 $ ] and by monotonicity @xmath677 either @xmath678 or @xmath679 . but
@xmath680 and @xmath681 so as @xmath233 is largeness preserving and @xmath682 is monotonic we can define @xmath683 .
this completes the proof that @xmath606 is a self - modulus .
we fill in the final piece of the puzzle by embellishing our construction from lemma [ lem : tree - minus - one ] so that the resulting downwardly generic tower is a uniquely small tower .
[ lem : tree - minus - one - small ] the statement of lemma [ lem : tree - minus - one ] still holds if we also demand that @xmath285 is largeness preserving and @xmath139 is uniquely @xmath62 small .
we sketch the modifications the proof of lemma [ lem : tree - minus - one ] requires . since it is trivial to ensure @xmath285 is largeness preserving simply by restricting which nodes we consider as values for @xmath684 we restrict our attention to ensuring that @xmath141 is uniquely @xmath62 small .
it is easy to eventually recognize pairs of nodes @xmath16 with @xmath685 in @xmath686 such that @xmath687{\beta+1}(x ) $ ] and to abandon ( remove from @xmath688 ) one or the other rendering it a terminal branch .
provided we always cut off one member of any such pair @xmath141 will surely be uniquely @xmath62 small .
the difficulty lies only in ensuring we choose the correct nodes to cut so as not to collaborate with our attempts to make @xmath141 eagerly generic in a way that prunes all infinite paths from @xmath141 .
our solution is to work in the domain rather than the image and regard @xmath689 to have priority @xmath12 .
note that we assume that @xmath690 implies that @xmath691 .
if at the end of stage @xmath295 we discover some minimal pair of strings @xmath16 and @xmath589 with @xmath692{\beta+1}(x ) $ ] where @xmath693 with @xmath694 we set @xmath689 to be undefined .
note that by working in the domain if we act to meet some genericity requirement by forcing all extensions of @xmath689 to pass through @xmath308 the only way @xmath308 could later be pruned from the tree is if @xmath689 is pruned from the tree so our additional pruning ca nt stop us from making @xmath141 eagerly generic .
we now argue that if @xmath695 then eventually @xmath689 settles down to a node that never gets pruned .
assume that @xmath19 is the node on @xmath138 with least code at which the claim fails , @xmath295 is a stage large enough that for every @xmath390 with with @xmath696 @xmath697 has settled on it s final value , @xmath19 never leaves @xmath138 after stage @xmath295 and that @xmath698 for all such @xmath390 .
let @xmath699 be larger than @xmath700 for all @xmath701 and @xmath702 the first stage following @xmath703 at which @xmath704 is reset .
if @xmath705 then @xmath706 so after @xmath707 @xmath708 is never again reset on account of avoiding simultaneous smallness and @xmath19 never leaves @xmath138 after @xmath295 so @xmath708 is nt reset after @xmath707 on account of the homeomorphism requirements so it must be reset at some @xmath709 on account of the genericity requirement .
but now nothing can reset @xmath708 if @xmath710 .
contradiction .
this enough to complete our proof of theorem [ thm : nonu ] .
invoke lemma [ lem : tree - minus - one - small ] to produce a a uniquely small tower @xmath644 with @xmath325 homeomorphic to @xmath11 with @xmath711 and set @xmath712 .
lemma [ lem : totally - generic ] tells us that @xmath52 is @xmath2-generic on @xmath21 and as @xmath52 is non - isolated lemma [ lem : generic - not - unif - sm ] guarantees that no @xmath610 is a uniform modulus for @xmath52 . on the other hand as both @xmath713 and @xmath714 are computable in @xmath281 so is @xmath52 and by lemma [ lem : singleton - or - jump ]
this entails that @xmath52 is a self - modulus . since every @xmath546 function has a moduli one
might also be prompted to ask whether there are @xmath546 degrees that are far from any modulus .
intuitively this should be true for sufficiently generic degrees and we verify this is the case .
[ thm : distant - modulus ] if @xmath52 is a non - isolated path through @xmath21 and @xmath52 is @xmath2 generic on @xmath21 _ relative _ to @xmath23 then no @xmath715 for some @xmath137 is a modulus for @xmath52 so in particular no @xmath716 is a modulus for @xmath52 . let @xmath717 $ ] be @xmath621 generic on @xmath21 relative to @xmath23 and let @xmath612 be a truth table functional , i.e. a turing functional total on all inputs .
suppose for a contradiction that @xmath718 , is a modulus for @xmath52 .
we work to build some @xmath719 not computing @xmath52 .
let @xmath720 and @xmath721 . at stage
@xmath722 we define @xmath723 such that @xmath724 and @xmath725 with @xmath726 .
thus at each stage we commit to some initial segment of @xmath51 and a function that @xmath51 must majorize
. our goal will be to force every turing reduction from @xmath51 either to disagree with @xmath52 or to be partial .
if there is a @xmath727 with @xmath728 forcing a disagreement between @xmath729 and @xmath52 then let @xmath730 and @xmath731 . otherwise if there is some @xmath732 with @xmath733 and an integer @xmath589 such that no @xmath727 with @xmath616 satisfies @xmath734 then leave @xmath735 and set @xmath736 .
if one of these two alternatives is satisfied for every @xmath26 then @xmath737 majorizes @xmath738 but fails to compute @xmath52 contradicting the assumption .
so suppose that for @xmath739 and @xmath740 neither alternative is satisfied .
we note that since @xmath52 is non - isolated , computably in @xmath741 one can enumerate an infinite list of distinct @xmath62 generic branches @xmath465 of @xmath21 with @xmath742 and @xmath743 with the later property guaranteed simply by letting @xmath465 equal @xmath52 on a long enough initial segment
. therefore we define @xmath744 so that @xmath745 searches for @xmath167 and @xmath404 with @xmath746 h^{g}(x)=y $ ] and then searches for values @xmath747 for each @xmath748 such that @xmath749 h^{g_i}(x)=y_i $ ] and returns a number larger than @xmath404 and all the @xmath747 .
such values @xmath747 must exist since @xmath750 is total .
thus @xmath751 , @xmath752 and @xmath753 .
since @xmath52 is @xmath754 generic relative to @xmath21 on @xmath21 there is some @xmath167 such that @xmath755 forces the @xmath67 where @xmath59 is the @xmath756{\beta } $ ] property that some extension @xmath757 with @xmath758 disagrees with @xmath52 .
fix @xmath589 so that @xmath759 and pick @xmath760 extending @xmath19 such that @xmath761 .
such a @xmath308 must exist as otherwise @xmath762 would have been a valid choice for @xmath763 .
moreover @xmath764 since @xmath52 forced @xmath765 and @xmath766 .
but as @xmath767 and @xmath768 we have @xmath769 and as @xmath770 is @xmath62 generic on @xmath21 some @xmath771 with @xmath772 forces @xmath59 contradicting the fact that @xmath755 forced @xmath765 .
the similarity between this result and lemma [ lem : generic - not - unif - sm ] is striking . since lemma [ lem : generic - not - unif - sm ] was in some sense a reflection of the fact that @xmath52 has a uniform modulus truth table computable within @xmath62 jumps if and only if @xmath52 is a @xmath565 singleton this naturally raises the following open question .
can the self - moduli be characterized in terms of definability like the characterization of the uniform self - moduli as the @xmath0 singletons ?
in particular we conjecture that @xmath52 is a self - modulus if and only if there is a computable tree @xmath21 with @xmath717 $ ] and a @xmath0 formula @xmath773 such that for all @xmath774 with @xmath775 $ ] the formula @xmath776 is uniquely satisfied by @xmath52 .
ideally , however , there would be a simpler statement expressing the definability of the self - moduli .
here we make good on our promise to define @xmath136 and @xmath148 .
[ def : notation - path ] a path from a limit notation @xmath34 to @xmath777 to is a sequence @xmath778 of notations such that @xmath779 and for every @xmath7 @xmath780 appears in the effective limit for @xmath781 , i.e. , @xmath782\exists({m})\left ( \gamma_{i+1 } = \kleenelim{\gamma_{i}}{m}\right)\ ] ] a path @xmath783 from @xmath34 to @xmath62 is minimal if @xmath784 for the least @xmath116 such that @xmath785 . given an initial segment @xmath540 of ordinal notations for every @xmath786 let @xmath148 denote the least limit notation @xmath146 in @xmath540 with @xmath787 for some @xmath116 .
if no such notation exists we write @xmath788 we first note that @xmath136 is always finite as otherwise the sequence @xmath798 would violate . to verify note that by applying the minimality of paths in @xmath540 we know that @xmath790{\beta } $ ] passes through @xmath799 for some @xmath800 .
claim now follows trivially by considering the paths from @xmath34 to @xmath35 .
it remains only to show the computability .
now we show if @xmath540 has a maximal element @xmath2 then @xmath136 is a computable function on @xmath540 . to this end
we define a decreasing sequence of ordinals @xmath801 dividing @xmath540 into connected pieces .
let @xmath802 and if @xmath801 is a limit ordinal set @xmath803 to be the predecessor of the least @xmath804 with @xmath805{\beta } $ ] defined .
if @xmath801 is a successor let @xmath803 be the predecessor of @xmath801 .
since this is a decreasing sequence of ordinals it must be finite thus for some @xmath26 @xmath806 and by definition [ def : nice - ordinal ] for every @xmath786 there is exactly one @xmath801 with @xmath805{\beta } $ ] defined .
we may compute @xmath136 by enumerating all paths from some @xmath801 until we find some path @xmath805{\beta } $ ] at which point we may set @xmath807{\beta } \rvert } $ ] .
if @xmath540 lacks a maximal element we note there is an increasing sequence @xmath808 cofinal in @xmath540 such that @xmath809 is nt defined for any @xmath7 .
by part we can build @xmath808 from any increasing cofinal sequence by repeatedly applying the operation taking @xmath62 to @xmath148 until no longer possible .
since the definition for @xmath136 in @xmath540 and @xmath810 agree when @xmath811 given @xmath2 we can simply compute @xmath136 on @xmath812 for an appropriate @xmath7 . to compute @xmath148 on @xmath810 we start listing @xmath813 and look for a @xmath34 and integer @xmath26 such that @xmath814 .
if such a pair is ever found we return @xmath815 .
simultaneously we start listing the sequence @xmath801 defined from @xmath816 and should we discover @xmath817 we return @xmath788 . the arguments given above guarantee that this is both a correct and complete procedure .
we start by enumerating @xmath825 and assigning @xmath826 the lower bound @xmath827 .
suppose we have already enumerated some @xmath828 with associated lower bound @xmath34 but have yet to enumerate any extension of @xmath822 .
here we search for the least @xmath26 such that @xmath829 .
let @xmath220 be the least notations such that @xmath830 for finite @xmath116 and @xmath831 and enumerate @xmath832 $ ] with lower bound @xmath34 .
otherwise let @xmath208 be the maximal notation with @xmath833 $ ] already enumerated .
now search for the least @xmath26 with @xmath834 and let @xmath220 be the least notations such that @xmath830 for finite @xmath116 and @xmath835 and enumerate @xmath832 $ ] with lower bound @xmath836 .
the construction clearly enumerates a sequence for every notation @xmath184 .
now assume that @xmath837 .
let @xmath838 be the longest common initial segment of @xmath839 and @xmath840 such that @xmath841 \subset { \beta^{\blacktriangleleft } } $ ] and @xmath842 \subset { \gamma^{\blacktriangleleft } } $ ] . during enumeration @xmath842
$ ] would have been tagged with a lower bound of at least @xmath843 .
hence @xmath844 but @xmath845 .
hence @xmath332 . to observe the other direction
note that if @xmath332 either @xmath837 or @xmath846 but the later possibility would entail that @xmath378 so @xmath837 .
given @xmath184 construct the sequence @xmath822 as by the prior lemma . by transfinite recursion
define the notation @xmath848 to be the successor of @xmath849 if @xmath62 is the successor of @xmath220 and the limit of @xmath850 where @xmath851 which is effective by the above construction .
now if @xmath852 then @xmath853 and hence for some @xmath781 @xmath854 $ ] so @xmath855 .
thus , @xmath856 .
moreover , note that there is a path from @xmath857 to @xmath858 only if @xmath34 appears in @xmath822 and that path is unique and minimal by construction satisfying part [ def : nice - ordinal : min ] of the definition .
if @xmath859 then @xmath860 so @xmath861 ensuring that part [ def : nice - ordinal : between ] of the definition is satisfied .
finally part [ def : nice - ordinal : finite - sequence ] is trivially satisfied as for each @xmath62 @xmath822 is a finite string . fix a unique path @xmath862 through @xmath31 and let @xmath808 be an increasing cofinal sequence in @xmath862 .
define @xmath863 and @xmath864 . by the definition of the effective addition operation on notations
there is no @xmath865 with @xmath866 and @xmath867 connected by a path but the set of ordinals therefore it follows from the fact that @xmath868 and @xmath801 are nice that @xmath803 is nice .
let @xmath869 . to simplify our notation slightly in the main body of the paper we ve made use of the fact that if @xmath148 is undefined then @xmath870 so we may safely set @xmath871 for any @xmath62 on which @xmath148 is undefined and by lemma [ lem : build - copylen ] can be done without imperiling the computability of @xmath136 .
note our construction of our nice path through @xmath31 provides a @xmath872 set of notations @xmath808 cofinal in @xmath31 such that the computations giving @xmath148 and @xmath136 for every @xmath873 can be uniformly computed from @xmath808 .
all constructions performed in the main body of the paper can be taken to use ordinals that lie along this @xmath872 path .
j. myhill .
finitely representable functions . in a.
heyting , editor , _ constructivity in mathematics : proceedings of the colloquium held at amsterdam , 1957 _ , studies in logic and the foundations of mathematics , pages 195207 .
north holland publishing co. , amsterdam , 1959 . | while much work has been done to characterize the turing degrees computing members of various collections of fast growing functions , much less has been done to characterize the rate of growth necessary to compute particular degrees .
prior work has shown that every degree computed by all sufficiently fast growing functions is uniformly computed by all sufficiently fast growing functions .
we show that the rate of growth sufficient for a function to uniformly compute a given turing degree can be separated by an arbitrary number of jumps from the rate of growth that suffices for a function to non - uniformly compute the degree .
these results use the unpublished method harrington developed to answer mclaughlin s conjecture so we begin the paper with a rigorous presentation of the approach harrington sketched in his handwritten notes on the conjecture .
we also provide proofs for the important computability theoretic results harrington noted were corollaries of this approach .
in particular we provide the first published proof of harrington s result that there is an effectively given sequence of @xmath0 singletons that are @xmath1 none of which is computable in the effective join of the @xmath2 jumps of the others for every @xmath3 .
= 1 |
consider the familiar linear regression model , @xmath3 where @xmath4 is an @xmath5-dimensional vector of responses , @xmath6 is the @xmath7 dimensional model matrix and @xmath8 is an @xmath5-dimensional vector of independent noise variables which are normally distributed , @xmath9 with variance @xmath10 .
let some arbitrary subset of the regression coefficients , @xmath11 , be zero meaning that the corresponding regressors do not contribute to the response in the underlying model .
the ordinary least squares ( ols ) estimate is obtained by minimizing the squared error loss .
although an unbiased estimator in this setting , the ols estimator may have large variance and will incorrectly estimate the coefficients that are zero in the underlying model . as a consequence , the estimator will usually result in overly complex models which may be difficult to interpret .
conventionally , an analyst will use subset selection to arrive at a reduced model , which is easier to interpret and attains better prediction accuracy .
the subset selection problem has been studied extensively @xcite . in cases with large numbers of variables ,
such methods suffer a fundamental limitation , the need for a greedy algorithm to search the discrete model space .
a more direct approach to improve the prediction accuracy of ols is based on `` shrinkage '' estimators @xcite which trade off increased estimator bias in return for a compensating decrease in variance leading to a smaller mean squared error ( mse ) .
a number of methods have been proposed that realize variable selection and estimation via some penalized least squares criteria , e.g. the nonnegative garotte @xcite , the lasso @xcite , the elastic net @xcite , the adaptive lasso @xcite and more recently the dantzig selector @xcite , etc . while traditional ridge regression @xcite proposes an @xmath12-penalty on the coefficients , the lasso , and its variants make use of the @xmath13-penalty .
@xcite demonstrates that the @xmath13-polytope , unlike @xmath12 , can touch the contours of the least squares objective function on one or more of the axes leading to estimates of zero for the associated regression coefficients .
@xcite synthesized several current ideas in the machine learning literature and offered two important hybrid algorithms for model selection and fitting in the kernel regression context .
in particular , by combining kernel transformations of independent variables with classical elements of bayesian hierarchical modeling @xcite he created the relevance vector machine ( rvm ) .
this approach is able to efficiently reduce the huge feature space created by the kernel transformations to a very parsimonious and predictive set improving significantly upon the less statistical support vector regression method of @xcite .
this continuous approach typically converges in a finite number of steps and provides very fast model selection for large number of variables without the concerns of monte - carlo noise or incomplete optimization typical of subset selection .
a number of extensions to the rvm have been offered ; see @xcite . recognizing the rvm as a bayesian random effects model , we offer an alternative formulation which offers a more complete hierarchical structure .
one major new development exploits this hierarchical structure to rapidly fit the model given a fixed value for the key hyper - parameter . unlike @xcite
, we do not integrate the regression coefficients out of the joint posterior distribution of the parameters .
instead , we use a conditional maximization procedure @xcite to obtain the posterior mode in an elegant and efficient way .
this approach to model fitting was criticized by ( * ? ? ?
* sec 8.3 ) in the context of standard linear mixed effects models due to the fact that it can lead to estimators of variance components that are identically zero and necessarily far from the posterior mean .
contrary to harville s conclusion , we show that the zeroing effect is theoretically justified and can be easily exploited for variable selection .
we refer to this procedure as the adaptive ridge selector ( aris ) . like the lasso , this results in a sparse shrinkage estimator which will zero irrelevant coefficients . the marginal likelihood @xmath14 is maximized over the hyper - parameter @xmath1 in order to select the best estimator .
this final empirical bayes step adjusts the amount of shrinkage imposed on the model . like rvm
, this algorithm typically converges in a finite number of steps and provides rapid and effective model selection and fitting for models with very large numbers of variables .
section [ hierarchical model ] introduces a hierarchical random effects model similar to that of @xcite and motivates our interest .
section [ optimize ] explains how the hierarchical structure is exploited to efficiently fit the model and describes the steps in the proposed aris algorithm .
it also analyzes the problem in the form of a regularized least squares problem and contrasts the estimator with one based on the marginal distribution of the regression coefficients .
next , section [ marginal ] focuses on deriving the marginal likelihood which is used to determine the optimal model . because the marginal likelihood is analytically intractable
, one must compute it through either analytical or simulation based approximation .
we provide a laplace approximation which is evaluated at the posterior mode along with a simulation based technique which results in somewhat more accurate solutions .
section [ examples ] presents comparisons of the proposed method along with alternatives on simulated data examples .
conclusions are discussed in section [ conclusions ] .
beginning with a standard hierarchical linear model ( * ? ? ?
* section 6.2.2 ) we propose a basic modification . in this case , the joint posterior of the parameters is proportional to , @xmath15 here a normal likelihood is assumed , @xmath16 , along with a conjugate normal prior on the regression coefficients , @xmath11 , and a typical jeffreys prior on the error variance @xmath17 , @xmath18 as with the relevance vector machine of @xcite , the vector @xmath19 where @xmath20 is a diagonal matrix with elements @xmath21 , @xmath22 the reciprocals of which are independent and identically distributed from a gamma distribution , @xmath23 where @xmath24 is the inverse scale parameter , and @xmath1 is the shape parameter . by definition @xmath25 , @xmath26 and @xmath27 .
notice that when @xmath28 , this becomes an exponential distribution which we will consider as a special case .
@xcite sets @xmath29 and @xmath30 which leads to a scale invariant improper prior .
he then derives a marginal likelihood @xmath31 through direct integration which is then maximized with respect to @xmath10 and @xmath32 .
hypothetically as the algorithm proceeds some @xmath21s will tend toward @xmath33 which correspond to the irrelevant variables in the model .
@xcite does not check the validity of the joint posterior density having assumed an improper prior on @xmath32 .
in contrast to @xcite , we choose not to integrate the regression coefficients out of the joint posterior distribution , but instead proceed to find the modal value given the data and @xmath1 . here
we fix @xmath24 to be a very small number ( e.g. machine epsilon ) and adjust @xmath1 to control shrinkage .
sparsity is obtained via the combination of these two particular priors , @xmath34 and @xmath35 .
integration over @xmath32 in the joint prior distribution @xmath36 will reveal that the marginal prior density of the regression coefficients is a product of univariate @xmath37 densities , with a scale of @xmath38 and degrees of freedom of @xmath39 .
it is important to note that the product of univariate @xmath37-densities is not equivalent to a multivariate @xmath37 and does not have elliptical contours but instead produces ridges along the axes .
these ridges can be made more drastic by choosing the scale parameter to be small ; see figure [ aris_fig1 ] . then the posterior will be maximized where ever these ridges first touch the contours of the likelihood .
the parameter @xmath1 plays a very similar role to the regularization parameter of the ridge regression , lasso , etc . , with larger values encouraging further shrinkage .
hence the proposed hierarchical structure implies independent @xmath37 priors being placed on each regression coefficient .
direct use of such @xmath37 priors would obscure the conjugate nature of the model . from an optimization perspective
, a direct use of such priors leads to a non - convex objective function which would not be desirable .
as we will discuss , within the hierarchical structure each iteration solves a simpler convex problem leading to an efficient solution .
once @xmath32s are integrated out of the joint posterior , the problem can be seen analogously as a regularized least squares problem as @xmath40 which solves @xmath41 having specified a complete hierarchical model , the joint posterior distribution of the parameters is obtained by the product of the likelihood and the specified priors up to a normalizing constant as @xmath42 where @xmath43 . given the priors in ( [ aris_eq2 ] ) , ( [ aris_eq3 ] ) , ( [ aris_eq4 ] ) , the product of these prior densities and the normal likelihood , ( [ aris_eq10 ] ) is the kernel of a posterior density function for @xmath11 , @xmath10 , and @xmath44 .
[ thm1 ] proof for this theorem can be found in appendix [ apppost ] .
the conditional distributions of the parameters can now easily be derived from this joint distribution . *
the regression coefficients are distributed as multivariate normal conditional on the error variance @xmath10 and the prior covariance of the regression coefficients @xmath20 .
@xmath45 where @xmath46 and @xmath47 . *
the error variance is distributed as inverse gamma conditional upon all other parameters .
@xmath48 thus , @xmath49 where @xmath50 and @xmath51 . * the prior precisions of the regression coefficients , conditional on all other parameters , follow a gamma distribution .
@xmath52 @xmath53 thus , @xmath54 where @xmath55 and @xmath56 . deriving the full set of conditional distributions
has several uses .
as is frequently done , we may utilize these to simulate from the posterior distribution using gibbs sampling .
such an approach would allow us to compute traditional bayes estimators for the regression coefficients . in section [ optimize ]
we show how to use the conditional distributions to maximize the joint posterior in a surprisingly simple and effective way .
maximization will also facilitate computation of the laplace approximation to the marginal likelihood @xcite .
@xcite proposed an optimization algorithm to find the joint posterior modes ; see also @xcite .
once the fully conditional densities of the model parameters are obtained , it is possible to maximize the joint posterior distribution by iteratively maximizing these conditional densities .
since the conditional posterior distributions obtained in equations ( [ aris_eq11 ] ) , ( [ aris_eq13 ] ) , and ( [ aris_eq15 ] ) are well - known distributions with readily available modes , the lindley - smith optimization algorithm becomes rather appealing to implement .
the modes for the distributions in equations ( [ aris_eq11 ] ) , ( [ aris_eq13 ] ) , and ( [ aris_eq15 ] ) respectively are @xmath57 where @xmath58 , @xmath59 were defined in ( [ aris_eq13 ] ) .
the maximization proceeds through sequential re - estimation of @xmath60 , @xmath61 , @xmath62 , where @xmath63 , @xmath64 is the number of iterations , and @xmath65 is the ols estimator . given @xmath20 , the modal value of @xmath11 can be obtained as a solution to a penalized least squares problem as with ridge regression . since we have an iterative procedure , let @xmath66 be the @xmath67th diagonal of @xmath68 at the @xmath69th iteration and be given .
then the @xmath69th iterate for @xmath11 is the solution to a similar penalized least squares problem : @xmath70 if we substitute @xmath21 with the estimate from ( [ aris_eq16 ] ) and let @xmath40 , we obtain @xmath71 where @xmath72 .
this procedure is essentially re - weighting the predictor variables by the positive square root of the ratio between the current estimate of the coefficients and the residual variance due to them .
after this re - weighting procedure , the problem takes the form of a standard ridge regression , @xmath73 where @xmath74 , @xmath75 and @xmath76 .
the solution to the problem above at iteration @xmath69 is given by @xmath77 hence the mode is computed through a sequence of re - weighted ridge estimators . the final estimate @xmath78 then can be recovered as @xmath79 ( this multiplication is understood component wise ) .
note that when @xmath80 , this procedure results in the ols estimator .
we construct a two - dimensional example to illustrate the method .
consider the model @xmath81 , where @xmath82 and @xmath83 .
we generate 30 observations and run aris for @xmath28 .
figure [ fig2 ] clearly demonstrates how the shrinkage proceeds throughout our algorithm .
the constrained region eventually becomes singular along the dimension which has no contribution to the response in the underlying model .
aris iteratively updates the constrained region and converges to a solution . ]
an expectation - maximization approach may be used to obtain the marginal posterior mode of @xmath11 .
consider the identity @xmath84 taking the logarithm and then taking the expectations of both sides with respect to @xmath85 yields @xmath86 where @xmath87 is the current guess of @xmath11 ( * ? ?
the em algorithm involves working with the first term of ( [ aris_eq40 ] ) .
the em procedure in our case would consist of the following two steps : ( i ) expectation of @xmath88 with respect to @xmath85 , ( ii ) maximization of the expected value with respect to @xmath11 .
an iterative procedure results by replacing the initial guess @xmath87 with the solution of the maximization procedure @xmath89 and repeating ( i ) and ( ii ) until convergence . with a slight change in the hierarchical model used , the above expectation will become quite trivial .
unlike ( [ aris_eq2 ] ) , let us not condition the prior density of @xmath11 on @xmath10 .
under such a setup , @xmath90 . notice that the conditional posteriors in ( [ aris_eq13 ] ) and ( [ aris_eq15 ] ) now become @xmath91 @xmath92 given the new prior , let us re - write ( [ aris_eq10 ] ) in the log form excluding the terms that do not depend on @xmath11 , @xmath10 and @xmath44 : @xmath93 next we compute @xmath94 as @xmath40 : @xmath95 where @xmath96 .
having completed the expectation step , the maximization of ( [ aris_eq43 ] ) with respect to @xmath11 yields an estimator as the solution to the following sequence of convex minimization problems : @xmath97 hence , the marginal posterior mode of @xmath11 is extremely similar in form to the joint posterior mode .
note that we can still adopt the weighting perspective mentioned earlier .
recall from section [ hierarchical model ] that the integration over @xmath32 in the prior distribution of @xmath98 results in a univariate @xmath37-density with degrees of freedom @xmath39 .
therefore , a value of @xmath99 will actually lead to a flat prior over @xmath98 resulting in the ols estimator ( note that when @xmath99 , the kernel of the @xmath37 density has power @xmath33 resulting in a flat density ) . also , the solution to the marginal when @xmath29 will be identical to the maximization of the joint posterior when @xmath28 .
@xcite mentions that the estimator of @xcite based on joint maximization may be far from the bayes estimator and suggests that the maximization of the marginal mode of the variance components would be a superior approach .
above we have shown that in our case the mode of the marginal density has the same form as the joint mode justifying the conditional maximization approach .
in fact , tipping adopts the approach suggested by harville , maximizing the joint posterior mode of @xmath44 and @xmath10 after integrating over @xmath11 but still achieves the zeroing effect .
we needed a slight change in the model to ease our work for the expectation step above , that is , we made the prior distribution of the regression coefficients independent of the error variance .
one may think that while we were trying to show the equivalence of these two solutions ( the joint and the marginal solutions ) , we actually created two different models and show that their solutions are identical in form , yet , they do not follow the same model . in the traditional bayesian analysis of the linear regression models ,
the regression coefficients are conditioned over the noise variance .
this provides an estimator for the regression coefficients that is independent of the noise variance .
we followed this convention when we were forming our hierarchical model .
however , in our case , there is no such thing as independence between the solutions of the regression coefficients and the noise variance .
although in an explicit statement such as ( [ aris_eq16c ] ) it may seem that the solution for the regression coefficients does not depend on the error variance , there exists an implicit dependence through the solution of @xmath32s .
we could have very well constructed our hierarchical model using a prior on regression coefficients independent of the noise variance .
this would lead to a solution that is only slightly different .
the re - estimation equation in ( [ aris_eq16c ] ) , ( [ aris_eq16b ] ) and ( [ aris_eq16 ] ) would become @xmath100 now , the solution for @xmath21 is independent of @xmath10 , yet the solution of @xmath11 explicitly depends on @xmath10 .
that said , the implicit dependence has become an explicit one .
thus the iterative solution for the regression coefficients as @xmath40 can be written as @xmath101 in which , apart from the tuning quantity ( @xmath102 ) , the only difference with ( [ aris_eq44 ] ) is the plug - in estimator used for the noise variance .
having shown that these procedures are fundamentally identical in form to each other , let us discuss another important point , the choice of initial values to start the algorithm .
let us consider the solution following ( [ aris_eq44 ] ) with @xmath29 : @xmath103\right)^{-1}\mathbf{x}'\mathbf{y } \label{aris_eq44b}\ ] ] we can not just plug any @xmath104 as an initial estimator . consider @xmath105 . in this case all the regression coefficients will be zeroed . or let only a subset of @xmath104 be zero . then in the solution those components will remain zero . although we are solving a series of simple convex problems , the dependency of the solution to the initial value proves that here we are dealing with a multi - modal objective function as would be expected . thus using the ols estimator as an initial value will take us to a local stationary point which is most likely under the support of the data in hand . to gain further intuition ,
let us consider an orthogonal case and let the predictors be scaled so that they have unit @xmath106-norm , i.e. @xmath107 .
in such a case the ols estimator for @xmath11 would have a variance - covariance matrix @xmath108 where @xmath109 is a plug - in estimator , e.g. the maximum likelihood estimator or the bias corrected estimator . testing the null hypothesis @xmath110 , a @xmath37-statistic can be computed for a component @xmath111 as @xmath112 .
notice in ( [ aris_eq44b ] ) the quantities at the diagonal of the second piece under the matrix inverse operation , @xmath113 , resemble the inverse of a squared @xmath37-statistic .
in fact , recall from section [ rrls ] that we formed a sequence of ridge regression problems out of this procedure by re - weighting our predictors by @xmath114 which is the absolute value of a @xmath37-statistic .
thus , following a conventional testing procedure , those predictors which correspond to coefficients with larger @xmath37-statistics will be given more importance .
this is yet another point that intuitively explains our procedure .
critical to the aris procedure is the choice of the hyper - parameter @xmath1 .
we propose an empirical bayes estimation of @xmath1 through the maximization of the marginal likelihood @xmath14 .
hence , we must integrate the joint posterior over all parameters , @xmath115 where @xmath116 . in the case of the hierarchical model developed in section [ hierarchical model ] ,
the direct calculation is intractable .
below we propose both analytical and simulation - based approximations .
a standard analytical approximation of the marginal likelihood can be computed using the laplacian method @xcite .
the approximation is obtained as @xmath117+\frac{p}{2}\log(2\pi)-\frac{1}{2}\log\left|\mathbf{h}_{\widetilde{\boldsymbol\theta}}\right| , \label{eq13d}\ ] ] where @xmath118 is the mode of the joint posterior and @xmath119 is the hessian matrix given in [ app ] evaluated at the posterior mode .
recall that the aris is designed to drive the values of @xmath98 and @xmath21 to zero for those independent variables @xmath120 which provide no explanatory value . as @xmath40 , the prior precisions and the regression coefficients related to irrelevant independent variables will tend toward @xmath121 and @xmath33 respectively .
in fact we can see in ( [ aris_eq21 ] ) that along these @xmath98 the curvature approaches @xmath121 as we converge to the solution thus driving their variance to 0 . at the joint posterior mode
, the corresponding dimensions of @xmath6 do not contribute and become irrelevant . under the support of the data , we claim these variables to be insignificant and suggest their removal from the model .
the integration follows removal of these irrelevant variables from the model .
the resulting laplace approximation to the log - marginal likelihood is @xmath122 where @xmath123 represents the reduced model after the removal of the irrelevant variables at the mode .
laplace approximation may not perform well in certain cases as will be seen in section [ examples ] .
@xmath11 and @xmath10 can be analytically integrated out of the joint posterior given in eq .
[ aris_eq10 ] .
the resulting likelihood conditioned on the prior variances is , @xmath124 where @xmath125 see also ( * ? ? ?
* equations 3.11,3.12 ) .
the marginal likelihood conditioned over @xmath1 can now be obtained through integration as @xmath126 $ ] where the expectation is taken over the prior distribution of @xmath44 . in order to ensure efficient sampling , we define a hypercube around the mode of the joint posterior in order to obtain a sampling region over @xmath127 .
the sampling region is the set @xmath128 where @xmath129 is the modal value of @xmath32 , @xmath130 is the square root of inverse curvature at the mode and @xmath131 is to be chosen to adjust the width of the box .
this section reports the results of a simulation study comparing the aris estimates with a number of computationally efficient penalized least squares methods . in the study
we consider a model of the form @xmath132 . for each data
set , we center @xmath4 and scale the columns of @xmath6 so that they have unit @xmath106-norm .
the lasso and elastic net were fit using the ` lars ` and ` elasticnet ` libraries in ` r ` .
_ model 0 _ : this model is adopted from @xcite and is a special case where the lasso estimate fails to improve asymptotically .
the true regression coefficients are @xmath133 .
the predictors @xmath134 ( @xmath135 ) are iid @xmath136 where @xmath137 is defined in @xcite ( corollary 1 , pg .
1420 ) with @xmath138 and @xmath139 . under this scenario
, @xmath137 does not allow consistent lasso selection . in this context
@xcite proposes the adaptive lasso ( adalasso ) for consistent model selection . in this setting
, we simulate 1000 data sets from the above model for different combinations of sample size and error variance .
table [ tab1 ] reports the proportion of the cases where the solution paths included the true model for aris , lasso and adaptive lasso .
we also report the results of aris in the special case when @xmath28 .
the results indicate that the aris algorithm performs nearly as well as the adaptive lasso and far better than the ordinary lasso in terms of consistent model selection under this particular setting . for @xmath28 ,
the aris produces a consistent estimate and does not require a search over the solution path . for medium and large values of @xmath5 we can see that it significantly outperforms the lasso .
results for the lasso and adalasso agree with those of @xcite .
we next compare prediction accuracy and model selection consistency using the following three models which are drawn from @xcite .
_ model 1 _ : in this example , we let @xmath140 with iid normal predictors @xmath134 ( @xmath135 ) . the pairwise correlation between the predictors @xmath120 and @xmath141 are adjusted to be @xmath142 .
_ model 2 _ : we use the same setup as model 1 with @xmath143 for all @xmath67 .
_ model 3 _ : we use the same setup as model 1 with @xmath144 .
we test models 1,2 , and 3 for two different sample sizes , @xmath145 and two noise levels @xmath146 .
this experiment is conducted 100 times under each setting . in table
[ tab2 ] , we report the median prediction error ( mse ) on a test set of 10,000 observation for each of the 100 cases .
the values in the parentheses give the bootstrap standard error of the median mse values obtained .
c , i and cm respectively stand for the number of correct predictors chosen , number of incorrect predictors chosen and the proportion of cases ( out of 100 ) where the correct model was found by the method .
the bootstrap standard error was calculated by generating 500 bootstrap samples from 100 mse values , finding the median mse for each case , and then calculating the standard error of the median mse .
lasso , adalasso , elastic net , nonnegative garrote , ridge and ordinary least squares estimates are computed along with the aris estimate . for the ridge estimator
, the ridge parameter is determined by a gcv ( generalized cross - validation ) type statistic , while for all the others we use 10-fold cross - validation for the choice of the tuning parameters .
we also consider the lasso where the tuning parameter is chosen by the method of @xcite .
aris hyper - parameter @xmath1 is determined both by the laplace approximation and the numerical integration to the marginal likelihood .
we also report the results for the particular case of @xmath28 . in each example the numerical integration step of aris - eb is carried out for values of @xmath147 and only the best result is reported .
this is a rather arbitrary choice and will depend upon the number of samples drawn .
model 3 is the only example where the same value of @xmath131 is consistently chosen ( @xmath148 ) .
.results for model 0 . [
cols="<,^,^,^",options="header " , ] [ tab2 ] model 3 demonstrates the most striking feature of the aris algorithm , the ability to identify the correct model under sparse setups . when utilized along with the empirical bayes step , it is able to identify the correct model in a very large proportion of cases with very low prediction error .
this is especially surprising for the cases where @xmath149 ( @xmath146 ) . in the case where @xmath150 and @xmath28 the algorithm still outperform all other methods in terms of correct model choice and mse . among all the variants of lasso ( lasso , adalasso , elastic net , lasso(cml ) ) , lasso(cml )
is optimal in terms of prediction accuracy . in the case of @xmath150 ,
its prediction error is almost identical to that of aris(@xmath28 ) but correct model identification is strongly weaker . observe that a cross - validation approach may not accurately choose the tuning parameter for the lasso - variants .
for example , as we moved from @xmath149 to @xmath150 , the proportion of cases where the correct model was chosen decreased for all the lasso - variants except lasso(cml ) where the tuning parameter is chosen via an empirical bayes step similar to our approach .
the nonnegative garrote estimator performs quite poorly in this situation along with the ridge and ols estimators .
results indicate that aris provides superior performance for model 3 . in the case of model 1 , for @xmath150 and @xmath151
, aris performs best in terms of prediction accuracy and strongly outperforms other algorithms in terms of model selection accuracy .
aris(@xmath28 ) outperformed other versions which required a search over the solution path .
both the laplace approximation and the numerical integration fail to detect this value of @xmath1 . for the case of @xmath149
, aris performs within a standard error of all the other estimators in terms of prediction accuracy , yet does better in terms of model selection accuracy .
the ridge estimator does almost as well as the lasso - variants in terms of prediction accuracy .
similar results follow for the case @xmath150 , @xmath152
. however , elastic net seems to have slightly lower prediction error .
the case @xmath149 , @xmath152 shows fairly weak results across all estimators .
model 2 demonstrates the biggest weakness of the aris and several other estimators .
when there are many small effects present in the underlying model , these estimators do not perform well since they favor parsimony . for all cases
the clear winner is the ridge estimator .
we have introduced a bayesian model fitting and variable selection method , aris , which makes use of a hierarchical model and enforces parsimony .
the method combines an efficient optimization procedure which is tailored to the fully conditional posterior densities with various techniques to derive and maximize the marginal likelihood . this development , although radically different in detail , is similar in spirit to modern implementation of the lasso which has been described as a bayesian procedure which combines a normal likelihood with a laplace prior on the regression coefficients ; see @xcite .
considering the simulation results of section [ examples ] , we note two key features of the aris : ( i ) its superior prediction and model selection accuracy when the underlying model is sparse , and ( ii ) the significant improvement in performance accompanying an increase in the sample size indicating asymptotic consistency .
computationally , for a specific @xmath1 value , aris requires one matrix inversion at each iteration .
this point is obvious from the description of the method as a series of ridge regressions .
thus the computational cost for each iteration of aris is at most @xmath153 . in practice , because variables are eliminated throughout the procedure , the cost often decreases dramatically with each iteration .
our experiments indicate fast convergence of this procedure across sample sizes .
lasso methods offer a computational advantage due to the _ lars _ algorithm @xcite which can compute the entire solution path of the lasso with the cost of a single ols estimator .
however , our experimental results indicate that these methods are often inferior in terms of model selection and prediction accuracy .
large scale experiments have shown that the procedure remains computationally feasible in situations where the number of predictor variables is very large .
hence the proposed method offers the most advantage in problems where one is attempting to select a small or moderate number of variables from a large initial group , a common situation in many modern statistical and data mining applications .
an open issue is the empirical bayes step via numerical integration . due to the large scale simulations throughout our experiments
we have only drawn @xmath154 samples for the integration of @xmath44 . obviously in practice a much larger set of samples could be drawn at little additional cost particularly for sparse models .
the process will become more stable as we draw larger samples .
in such a case , the choice of @xmath131 may just be fixed at a larger value , i.e. @xmath148 .
the authors would like to express their appreciation to robert mee and william m. briggs for their suggestions which have significantly added to the clarity of the manuscript .
@xmath11 and @xmath10 can tractably be integrated out of ( [ aris_eq10 ] ) . as a result of this integration ,
the only remaining terms that are dependent upon @xmath32 are @xmath155 it will suffice to show that ( [ app - eq1 ] ) is finitely integrable with respect to @xmath32 .
@xmath156 and @xmath157 eliminating the terms again that are not dependent upon @xmath21 , we reduce ( [ app - eq1 ] ) to @xmath158 integrating ( [ app - eq2 ] ) is equivalent to @xmath159 .
this expectation is taken with respect to ( [ aris_eq4 ] ) and is finite for @xmath26 .
@xmath162 \label{aris_eq25 } \\
-\frac{\partial^{2}}{\partial\sigma^{2}\partial v_{k}^{-1}}\log p\left(\mathbf{y},\boldsymbol\theta|\eta\right ) & = & -\frac{\beta_{k}^{2}}{2\sigma^{4}}. \label{aris_eq26}\end{aligned}\ ] ] drucker , h. , burges , c. j. c. , kaufman , l. , smola , a. , and vapnik , v. `` support vector regression machines . '' in mozer , m. c. , jordan , m. i. , and petsche , t. ( eds . ) , _ advances in neural information processing systems _ , volume 9 , 155 . the mit press ( 1997 ) .
dsouza , a. , vijayakumar , s. , and schaal , s. `` the bayesian backfitting relevance vector machine . '' in _
icml 04 : proceedings of the twenty - first international conference on machine learning _ , 31 ( 2004 ) . | we introduce a new shrinkage variable selection operator for linear models which we term the _ adaptive ridge selector _ ( aris ) .
this approach is inspired by the _ relevance vector machine _ ( rvm ) , which uses a bayesian hierarchical linear setup to do variable selection and model estimation . extending the rvm algorithm
, we include a proper prior distribution for the precisions of the regression coefficients , @xmath0 , where @xmath1 is a scalar hyperparameter .
a novel fitting approach which utilizes the full set of posterior conditional distributions is applied to maximize the joint posterior distribution @xmath2 given the value of the hyper - parameter @xmath1 .
an empirical bayes method is proposed for choosing @xmath1 .
this approach is contrasted with other regularized least squares estimators including the lasso , its variants , nonnegative garrote and ordinary ridge regression .
performance differences are explored for various simulated data examples .
results indicate superior prediction and model selection accuracy under sparse setups and drastic improvement in accuracy of model choice with increasing sample size .
* artin armagan * + .05 in department of statistics , operations , and management science + the university of tennessee + knoxville , tn , 37996 + _ e - mail : _ [email protected] .1 in * and * + .1 in * russell l. zaretzki * + .05 in department of statistics , operations , and management science + the university of tennessee + knoxville , tn , 37996 + _ e - mail : _ [email protected] + .1 in keywords : lasso ; elastic net ; shrinkage estimation ; rvm ; ridge regression ; lars algorithm ; penalized least squares . |
a point - like peak in the primordial density of the universe results , well after recombination and decoupling , in two spherical shells of overdensity : one of radius @xmath8 mpc , and one of radius @xmath9 mpc @xcite .
( all distances in this article are co - moving " , i.e. are referred to the present time @xmath10 . )
the acoustic length scale " @xmath11 mpc is approximately the distance that sound waves of the tightly coupled plasma of photons , electrons , protons , and helium nuclei traveled from the time of the big bang until the electrons recombined with the protons and helium nuclei to form neutral atoms , and the photons decoupled .
the inner spherical shell of @xmath9 mpc becomes re - processed by the hierarchical formation of galaxies @xcite , while the radius of the outer shell is unprocessed to better than 1% @xcite ( or even 0.1% with corrections @xcite ) and therefore is an excellent standard ruler to measure the expansion of the universe as a function of redshift @xmath3 .
histograms of galaxy - galaxy distances show an excess of galaxy pairs with distances in the approximate range @xmath12 to @xmath13 mpc . this
baryon acoustic oscillation " ( bao ) signal has a signal - to - background ratio of the order of 0.1% to 1% .
measurements of these bao signals are by now well established ( see @xcite for extensive lists of early publications ) . in this article
we present studies of bao with sloan digital sky survey ( sdss ) data release dr12 @xcite .
to establish the notation , let us recall the equations that describe the metric and evolution of a homogeneous universe in the general theory of relativity : @xcite @xmath14 , \nonumber \\
\label{s } \\
\frac{h}{h_0 } & \equiv & e(a ) = \sqrt{\frac{\omega_\textrm{m}}{a^3 } + \frac{\omega_{\textrm{r}}}{a^4 } + \omega_\textrm{de}(a ) + \frac{\omega_k}{a^2 } } , \label{e}\end{aligned}\ ] ] where the expansion and hubble parameters are defined as @xmath15 we rescale @xmath16 such that the curvature constant @xmath17 adopts one of three values : @xmath18 or @xmath19 for a spatially flat , closed or open universe , respectively . today @xmath20 ,
@xmath21 and @xmath22 .
the red shift " @xmath3 is related to the expansion parameter : @xmath23 where @xmath24 and @xmath25 are the frequencies of emission and reception of a ray of light as measured by comoving observers .
note that along a light ray @xmath26 is constant .
@xmath27 and @xmath28 are , respectively , the densities of non - relativistic matter and ultra - relativistic radiation measured in units of the critical density " @xmath29 .
@xmath30 ^ 2 $ ] . in the general theory of relativity @xmath31 , where @xmath32 is the cosmological constant " . to test the theory
we let @xmath33 ( with @xmath34 ) be a function of @xmath7 to be determined by observations .
note that from eq .
( [ e ] ) at @xmath10 we obtain the identity @xmath35 it is observed that @xmath36 @xcite so we expand expressions up to first order in @xmath37 . at the late times of interest the density of radiation is negligible with respect to the density of matter so we set @xmath38 .
we consider four scenarios : 1 . the observed acceleration of the expansion of the universe is due to the cosmological constant , i.e. @xmath39 .
the observed acceleration of the expansion of the universe is due to a gas of negative pressure with an equation of state @xmath40 .
we allow the index @xmath41 to be a function of @xmath7 @xcite : @xmath42 . while this gas dominates @xmath43 the equation @xcite @xmath44 can be integrated with the result @xcite @xmath45 if @xmath46 and @xmath47 we obtain @xmath39 as in the general theory of relativity .
3 . same as scenario 2 with @xmath48 constant , i.e. @xmath47 .
we measure the dependence of @xmath6 on @xmath7 in the linear approximation @xmath49 note that bao measurements can constrain @xmath6 for @xmath50 where @xmath6 contributes significantly to @xmath43 .
problem : an astronomer observes two galaxies at the same redshift @xmath3 separated by an angle @xmath51 . what is the distance @xmath52 between these two galaxies today ?
answer to sufficient accuracy : @xmath53 here we have written the speed of light @xmath54 explicitly .
the function @xmath55 is presented in table [ chi ] for some bench mark universes .
( @xmath55 is not to be confused with the @xmath56 of fits . )
.[chi ] function @xmath55 for several values of @xmath3 , @xmath57 and @xmath58 .
measurements at the lyman alpha forest " correspond to @xmath60 .
decoupling of photons occurs at @xmath61 @xcite .
all columns have @xmath46 , @xmath47 , except the row ( * ) which has @xmath62 , @xmath47 .
an approximation to @xmath55 good to an accuracy of @xmath63 for @xmath3 in the range 0 to 1 is @xmath64 with @xmath65 given in the next - to - last row .
@xmath66 is calculated from eq .
( [ thetamc_rs ] ) with @xmath4 given by ( [ tmc ] ) . [ cols="^,^,^,^,^,^ " , ]
let us try to understand qualitatively how the bao distance measurements presented in table [ d0 ] constrain the cosmological parameters .
in the limit @xmath67 we obtain @xmath68 , so the row with @xmath69 in table [ d0 ] approximately determines @xmath70 .
this @xmath70 and the measurement of , for example , @xmath71 then constrains the derivative of @xmath72 with respect to @xmath7 at @xmath73 , i.e. constrains approximately @xmath74 or equivalently @xmath75 .
the fit to the bao data of table [ d0 ] for scenario 1 ( with the correction ( [ dd ] ) with @xmath76 ) obtains @xmath77 , while the constraint on the orthogonal relation is quite weak : @xmath78 .
we need an additional constraint for scenario 1 . at small @xmath7 , @xmath43
is dominated by @xmath57 , so @xmath4 plus @xmath70 approximately constrain @xmath57 , or equivalently @xmath79 , see eq .
( [ rs2 ] ) .
the additional bao distance measurements in table [ d0 ] then also constrain @xmath80 and @xmath81 or @xmath82 . in table
[ bao_fit ] we present the cosmological parameters obtained by minimizing the @xmath56 with 18 terms corresponding to the 18 bao distance measurements in table [ d0 ] for several scenarios .
we find that the data ( with the correction ( [ dd ] ) with @xmath76 ) is in agreement with the simplest cosmology with @xmath83 and @xmath6 constant with @xmath56 per degree of freedom ( d.f . )
@xmath84 , so no additional parameter is needed to obtain a good fit to this data .
note that the bao data alone place a tight constraint on @xmath77 for constant @xmath6 , or @xmath85 when @xmath6 is allowed to depend on @xmath7 as in scenario 4 .
the constraint on @xmath37 is weak . in table
[ bao_thetamc_fit_12 ] we present the cosmological parameters obtained by minimizing the @xmath56 with 19 terms corresponding to the 18 bao distance measurements listed in table [ d0 ] ( with the correction ( [ dd ] ) with @xmath76 ) plus the measurement of @xmath4 from the cmb given in eq .
( [ tmc ] ) .
we present the variable @xmath86 instead of @xmath58 because it has a smaller uncertainty .
the simplest cosmology with @xmath83 and @xmath6 constant has @xmath56 per d.f . @xmath87 and is therefore disfavored with a significance corresponding to @xmath5 for one degree of freedom .
releasing one of the parameters @xmath37 or @xmath80 or @xmath82 obtains a good fit to the data . releasing only @xmath37 obtains @xmath88 with @xmath56 per d.f .
@xmath89 . releasing only @xmath80 obtains @xmath90 with @xmath91 . releasing only @xmath82 obtains @xmath92 with @xmath93 . releasing both @xmath37 and
@xmath82 obtains @xmath56 per d.f .
@xmath94 , @xmath95 and @xmath96 .
details of these and other fits are presented in table [ bao_thetamc_fit_12 ] .
table [ bao_thetamc_fit_00 ] presents the corresponding fits with no correction for peculiar motions . in summary , we find that the bao@xmath97 data are in agreement with a family of universes with different @xmath37 .
two examples are presented in table [ bao_thetamc_fit_12 ] : see the fits for scenario 4 .
fixing @xmath98 we obtain @xmath58 independent of @xmath7 within uncertainties : @xmath99 .
fixing @xmath83 we obtain @xmath6 that does depend significantly on @xmath7 : @xmath92
. however , if we require both @xmath83 and @xmath6 constant we obtain disagreement with the data with @xmath56 per d.f .
the reference fit in this section is scenario 1 with @xmath83 fixed , @xmath56 with 19 terms : 18 bao terms from table [ d0 ] ( corrected with @xmath76 ) plus @xmath4 from eq .
( [ tmc ] ) .
the @xmath56 per d.f . is @xmath87 , see table [ bao_thetamc_fit_12 ] .
this tension is equivalent to @xmath5 for one degree of freedom .
releasing a single parameter reduces the tension to less than @xmath100 .
the main contributions to the @xmath56 of the reference fit are @xmath101 contributing 8.5 , @xmath102 , @xmath103 , @xmath104 , @xmath105 , and @xmath106 contributing 4.7 .
so we see no obvious pattern or mistake in the identification of the bao signals .
removing the term corresponding to @xmath4 from the @xmath56 obtains @xmath84 , so the tension is between the @xmath4 measurement and the bao measurements . does @xmath4
need a correction ? here
are fits with less bao terms in the @xmath56 than the reference fit .
fitting the 6 @xmath1 measurements plus @xmath4 obtains @xmath107 . fitting the 6 @xmath0 measurements plus
@xmath4 obtains @xmath108 . fitting the 6 @xmath2 measurements plus
@xmath4 obtains @xmath109 . removing @xmath104 obtains @xmath110 .
removing the 3 bao entries with @xmath69 obtains @xmath111 .
removing the 3 bao entries with @xmath112 obtains @xmath113 .
removing the 6 bao entries with @xmath69 and @xmath114 obtains @xmath115 .
removing the 6 bao entries with @xmath116 and @xmath112 obtains @xmath117 . in conclusion
, none of these removals of bao measurements from the @xmath56 of the fit removes the tension .
we now perform a fit with different bao measurements listed in table [ d1 ] : @xmath118 g - c , @xmath119 g - c , @xmath120 g - lc , and @xmath121 lg - lg .
these measurements have different selections of centers and/or galaxies , or different binnings of @xmath3 , and different fits from the measurements in table [ d0 ] .
the fit to these 12 bao measurements plus @xmath4 for scenario 1 with @xmath83 fixed obtains @xmath122 , equivalent to a tension of @xmath123 for 1 degree of freedom .
releasing only @xmath37 obtains @xmath124 with @xmath125 .
releasing only @xmath80 obtains @xmath126 with @xmath127 . releasing only @xmath82 obtains @xmath128 with @xmath129 . note that releasing a single parameter removes the tension .
these last three fits are consistent with the ones presented in table [ bao_thetamc_fit_12 ] , and have @xmath56 per d.f . less than 1 confirming that we are not underestimating the systematic uncertainty of the bao distances .
to remove the tension between the scenario with @xmath83 fixed and @xmath6 constant it is necessary to shift @xmath4 by @xmath130 or increase the systematic uncertainty of the bao distances
@xmath131 by a factor 1.7 .
it is difficult to imagine that the systematic uncertainty is wrong by a factor 1.7 when it was obtained directly from the data , and in addition we obtain @xmath56 per d.f .
close to 1 by releasing a single parameter .
the conclusions are : ( i ) scenario 1 with @xmath83 and constant @xmath6 has tension with the bao plus @xmath4 data , and ( ii ) dropping the constraint @xmath83 and/or allowing @xmath6 to be variable leaves no significant tension .
these conclusions are robust with respect to the selection of the bao data .
we obtain @xmath6 from the 6 independent measurements of @xmath1 in table [ d0 ] with the correction @xmath76 and eqs .
( [ dz_rs ] ) and ( [ e ] ) , for the cases @xmath98 and @xmath83 .
the values of @xmath70 and @xmath58 are obtained from the fits for scenario 4 in table [ bao_thetamc_fit_12 ] .
the results are presented in fig .
[ o_de_z ] . to guide the eye
, we also show the straight line corresponding to scenario 4 . to cross - check the robustness of @xmath6 in fig .
[ o_de_z ] we add the 17 measurements of @xmath1 in table [ d1 ] and obtain fig .
[ o_de_z_all ] . note that these measurements of @xmath1 are partially correlated .
note that @xmath6 is consistent with a constant for @xmath98 , but not for @xmath83 .
the main results of these studies are the 18 independent bao distance measurements presented in table [ d0 ] which do not depend on any cosmological parameter .
these bao distance measurements alone place a strong constraint on @xmath77 ( for constant @xmath58 ) or @xmath85 ( when @xmath6 is allowed to depend on @xmath7 ) , while the constraint on @xmath37 is weak .
constraints on the cosmological parameters in several scenarios from the bao measurements alone and from bao plus @xmath4 measurements are presented in tables [ bao_fit ] , [ bao_thetamc_fit_12 ] and [ bao_thetamc_fit_00 ] .
we find that the 18 bao distance measurements plus @xmath132 are in agreement with a family of universes with different @xmath37 .
two examples are @xmath98 with @xmath6 constant , and @xmath83 with @xmath6 varying significantly with @xmath7 .
for these two examples we present measurements of @xmath6 in figs .
[ o_de_z ] and [ o_de_z_all ] .
the cosmology with both @xmath83 and @xmath6 constant has a tension of @xmath5 with the bao plus @xmath4 data .
the bao plus @xmath4 data for constant @xmath6 obtains @xmath88 for @xmath76 and @xmath133 for @xmath134 .
these results have some tension with independent observations : @xmath135 from cmb data alone with the assumption of constant @xmath6 @xcite , and @xmath136 from cmb plus supernova ( sn ) data with the assumption of constant @xmath6 @xcite .
if @xmath6 is allowed to be variable there are only loose constraints on @xmath37 from independent measurements , so the case with @xmath83 with non - constant @xmath6 is viable .
funding for sdss - iii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , and the u.s .
department of energy office of science .
the sdss - iii web site is http://www.sdss3.org/. sdss - iii is managed by the astrophysical research consortium for the participating institutions of the sdss - iii collaboration including the university of arizona , the brazilian participation group , brookhaven national laboratory , carnegie mellon university , university of florida , the french participation group , the german participation group , harvard university , the instituto de astrofisica de canarias , the michigan state / notre dame / jina participation group , johns hopkins university , lawrence berkeley national laboratory , max planck institute for astrophysics , max planck institute for extraterrestrial physics , new mexico state university , new york university , ohio state university , pennsylvania state university , university of portsmouth , princeton university , the spanish participation group , university of tokyo , university of utah , vanderbilt university , university of virginia , university of washington , and yale university . | we define baryon acoustic oscillation ( bao ) distances @xmath0 , @xmath1 , and @xmath2 that do not depend on cosmological parameters .
these bao distances are measured as a function of redshift @xmath3 with the sloan digital sky survey ( sdss ) data release dr12 . from these bao distances alone , or together with the correlation angle @xmath4 of the cosmic microwave background ( cmb ) , we constrain the cosmological parameters in several scenarios .
we find @xmath5 tension between the bao plus @xmath4 data and a cosmology with flat space and constant dark energy density @xmath6 .
releasing one and/or the other of these constraints obtains agreement with the data .
we measure @xmath6 as a function of @xmath7 . |
quantum entanglement is one of the central concepts in quantum information theory , and can be viewed as a physical resource for quantum information@xcite . characterizing entanglement is still an open question for many - body system , however , for two - qubit case , the problem is solved by using concurrence which is defined by wootters@xcite . in refs.amico,horodecki,wineland,kitagawa,toth,alekseev,soensen,pu,messikh,wang2009,yin,orgikh , it was found that spin squeezing has a close relation with entanglement , and spin squeezing parameters can be used as entanglement witnesses .
the definitions of spin squeezing parameters are not unique , and the most popular parameters are @xmath0 , given by kitagawa and ueda@xcite , and @xmath1 , given by wineland _ _ et al__@xcite .
the parameter @xmath0 was shown to be related to pairwise correlations and concurrence____kitagawa , orgikh , messikh , wang2009,yin , while the parameter @xmath1 is related to many - body entanglement____wineland , messikh .
there is another spin squeezing parameter proposed by srensen _ _ et al__@xcite as a measure of many - body entanglement which can be considered as a generalization of parameter @xmath2____@xcite .
recently , a new spin squeezing parameter @xmath3 was proposed , and it was found that if @xmath4 , the system is entangled____@xcite .
reference @xcite found that spin squeezing parameter @xmath0 has close relation with pairwise correlation . here , we consider the following pairwise correlation @xmath5 in the @xmath6 direction , where @xmath7 , with @xmath8 being the pauli matrix of the @xmath9-th qubit and @xmath6 being the normalized direction vector .
the above definition can be rewritten as @xmath10where @xmath11 is the pairwise correlation matrix whose elements are given by @xmath12 in ref .
@xcite , it was found that , if the system has exchange symmetry , the pairwise correlation and the spin squeezing parameter @xmath13 have the following relation,@xmath14where indices @xmath9 , @xmath15 were omitted due to exchange symmetry , @xmath16 is the direction which is perpendicular to the mean spin direction @xmath6 , and @xmath17 is the number of qubits . in this paper , we find that the parameter @xmath3 is related to the minimal pairwise correlation for systems with only symmetric dicke states populated @xcite .
the relation is given by @xmath18where the minimization is over arbitrary direction @xmath6 .
thus , for systems with only symmetric dicke states populated , negative correlations @xmath19 are equivalent to the spin squeezing @xmath20 .
many works have been devoted to understand spin squeezing , quantum entanglement , and quantum chaos____gorin , jacquod , wang , wang0,wang2003,weinstein0,ghose , stamatiou , song , furuya , petijean , soensen , song0 . here
, quantum chaos@xcite mainly focuses on the researches of quantum characteristics of a quantum system whose classical correspondence exhibits chaos . in refs.____song ,
song0 , the authors used the parameter @xmath0 as an efficient signature of quantum chaos .
they also studied the relations between @xmath13 and concurrence , and found that @xmath0 does not match concurrence well . in this paper
, we use the new spin squeezing parameter @xmath21 to characterize quantum chaos , and study the dynamical evolutions of concurrence and spin squeezing parameters @xmath0 and @xmath21 .
the quantum kicked top ( qkt ) model____arecchi , haake , ariano , ghose , stamatiou , song is a typical model that exhibits quantum chaos , and its chaotic behaviors were demonstrated experimentally by using linear entropy____@xcite . in this paper
we find that the concurrence decreases abruptly and non - smoothly to zero in a finite time in the qkt model , and this phenomenon is called entanglement sudden death ( esd)@xcite , which has been widely studied both theoreticallyyu2004,yu2006,cavalcanti , huang , almeida , sainz , qasimi , yamanoto , salles , yu2009,faria , pablo , weinstein2009 and experimentally____@xcite in decoherence dynamics . similar to the definition of esd , there is entanglement sudden birth ( esb)@xcite , which is a sudden feature in the temporal creation of entanglement in a dissipative evolution of interacting qubits . in ref.@xcite , the authors found that spin squeezing sudden death ( sssd ) may occur due to decoherence . in this paper , we want to study these sudden features of entanglement and spin squeezing in the qkt model .
this paper is organized as follows . in sec.____ii
, we first introduce the definitions of spin squeezing parameters , negative correlations , and concurrence , then give the relationship between the spin squeezing parameter @xmath3 and the minimal pairwise correlation @xmath22 .
in sec.____iii , we introduce the qkt model and its classical correspondence . in sec.____iv
, we analyze the quantum chaos of the qkt model by means of the dynamics of spin squeezing and entanglement . in sec.____v , we study the influences of chaos on esd , esb , sssd and spin squeezing sudden birth ( sssb ) .
the conclusions are given in sec.vi .
the spin squeezing parameters are useful tools to detect the quantum entanglement @xcite . here
, we consider an ensemble of @xmath17 spin-@xmath23 particles described by the collective angular momentum operators @xmath24the dicke states can be expressed as @xmath25 , where @xmath26 is the spin down state and @xmath27 .
we mainly study the following two types of spin squeezing parameters @xmath13 and @xmath3 .
the parameter @xmath0 is defined as @xcite @xmath28where @xmath29 denotes the direction which is perpendicular to the mean spin direction @xmath30 , and @xmath31 is the minimal value of the variance @xmath32 in the @xmath33-direction .
the spin squeezing parameter @xmath0 can be written in an explicit form as@xcite@xmath34_{+}\rangle ^{2}}\bigg { ] } , \label{sp_ku1}\end{aligned}\]]where @xmath35 and @xmath36 are two orthogonal directions which are perpendicular to the mean spin direction , and @xmath37 _ { + } = ab+ba$ ] .
the spin squeezing parameter @xmath3 is defined as toth , wang2009@xmath38where @xmath39 is the minimal eigenvalue of the real symmetric matrix @xmath40where @xmath41 is the correlation matrix of which the matrix elements are @xmath42the covariance matrix @xmath43 is given as@xmath44 below , for the sake of simplicity , we assume the mean spin direction @xmath45 is along the @xmath46-axis .
and then we can write the matrix @xmath47 in an explicit form as@xmath48where @xmath49 is a @xmath50 matrix@xmath51 _ { + } \big\rangle \\
\frac{n}{2}\big\langle \left [ j_{x},j_{y}\right ] _ { + } \big\rangle & n% \big\langle j_{y}^{2}\big\rangle% \end{array}% \right ) , \]]and @xmath52 _ { + } % \big\rangle , \frac{n}{2}\big\langle \left [ j_{z},j_{y}\right ] _ { + } % \big\rangle \right ) .\end{aligned}\]]note that @xmath53 is a @xmath54 vector . according to eq.(sp_ku1 ) , the parameter @xmath0 is equal to the minimal eigenvalue of the matrix@xmath55 _ { + } \big\rangle \\
\frac{n}{2}\big\langle \left [ j_{x},j_{y}\right ] _ { + } \big\rangle & n% \big\langle j_{y}^{2}\big\rangle% \end{array}% \right ) .\]]in analogy to the relation between @xmath0 and @xmath56 , we define a spin squeezing parameter@xmath57 ,
\label{zetann}\]]which characterizes the spin squeezing along the mean spin direction .
here we consider the case of @xmath58 . according to the rayleigh - ritz theorem@xcite ,
the minimal eigenvalue of @xmath47 is less or equal to that of @xmath49 , thus we have @xmath59 .
now , we discuss the relation between the spin squeezing parameter @xmath60 and the minimal pairwise correlation @xmath61 . here
, we consider the system with only symmetric dicke states populated , which has exchange symmetry and @xmath62 . in this case
, we have the following relations , @xmath63 _ { + } \big\rangle & = \frac{% n\left ( n-1\right ) } { 4}\big\langle \left [ \sigma _ { 1\alpha } , \sigma _ { 2\beta } \right ] _ { + } \big\rangle , \text { } \left ( \alpha \neq \beta \right).\end{aligned}\]]thus , with the above relations , the matrix @xmath47 , as shown in eq.([pc - sp ] ) , can be rewritten as@xmath64 _ { + } \big\rangle -\left ( n-1\right ) \big\langle j_{\alpha } \big\rangle \big\langle j_{\beta } \big\rangle \notag \\ & = \frac{n^{2}\left ( n-1\right ) } { 4}\left ( \big\langle \sigma _ { \alpha } \sigma _ { \beta } \big\rangle -\big\langle \sigma _ { \alpha } \big\rangle % \big\langle \sigma _ { \beta } \big\rangle \right ) \notag \\ & = \frac{n^{2}\left ( n-1\right ) } { 4}\mathbb{c}_{\alpha , \beta } , \text { \ \ } % \left ( \alpha \neq \beta \right ) ; \end{aligned}\]]@xmath65the relation between the matrix @xmath47 and the pairwise correlation matrix @xmath11 can be written as@xmath66where @xmath67 is a @xmath68 identity matrix .
thus , the matrix @xmath69 and @xmath47 can be diagonalized with the same unitary transformation .
so @xmath70where @xmath61 and @xmath39 are the minimal eigenvalues of the matrix @xmath11 and @xmath47 , respectively . here
the spin squeezing parameter @xmath3 can be written as @xmath71 since @xmath72 .
the relation between the minimal pairwise correlation and the parameter @xmath3 can be written as @xmath73therefore , negative correlation ( @xmath74 ) is equivalent to spin squeezing ( @xmath75 ) . from refs.____toth , wang2009
, @xmath75 is also a criterion of entanglement , so it reveals that there are close relations among negative correlation , spin squeezing , and entanglement .
if the minimal pairwise correlation is in the plane which is perpendicular to the mean spin direction , the above relation will degenerate to eq.([pc_ku ] ) . here
, we briefly introduce the relations between spin squeezing and concurrence .
the entanglement between a pair of spin-@xmath23 particles is quantified by the concurrence @xmath76@xcite , which is defined as@xmath77where the quantities @xmath78 are the square roots of the eigenvalues of the matrix @xmath79 in descending order , @xmath80 is the complex conjugate of @xmath81 . for spin states with parity
, it was found that@xcite , when @xmath82 , the relation between @xmath0 and @xmath76 is @xmath83 in refs.@xcite , the authors found that for systems with parity and exchange symmetry , when concurrence @xmath84 , the spin squeezing parameter @xmath75 , and vice versa .
now we introduce the qkt model .
consider an ensemble of @xmath17 spin-@xmath23 particles , the qkt hamiltonian reads@xcite @xmath85where @xmath86 , and @xmath87 @xmath88 are angular momentum operators that obey the commutation relations @xmath89
= i\hbar \varepsilon _ { \alpha \beta \gamma } j_{\gamma } $ ] , where @xmath90 is the levi - civita symbol . the first term of eq.([qkt_hamiltonian ] ) describes a nonlinear precession around the @xmath46-axis with strength @xmath91 , and the second term describes the kicks around the @xmath92-axis with strength @xmath93 , separated by a period @xmath94 . in the below , we set @xmath95 and @xmath96 , and the magnitude @xmath97 is a constant of the motion .
now , we study the classical corresponding of the qkt .
the evolutions of the expectation values of the angular momentum operators are@xmath98where @xmath99 is the floquet operator describing the unitary evolution for each kick,@xmath100to study the quantum chaos of the qkt , we should first analyze its classical corresponding , which is obtained in the classical limit , i.e. @xmath101 . for convenience , we use the following three quantities @xmath102when @xmath103 , these three variables become@xmath104where @xmath105 is the polar angle and @xmath106 is the azimuthal angle .
therefore , @xmath107 represents a point on the bloch sphere with radius @xmath108 . in the classical limit
, we can factorize the products of the mean values of the angular momentum operators as @xmath109by substituting eqs.([floquet ] ) , ( [ jxyz ] ) , and ( [ jxy ] ) into eq.([evolution ] ) , we can derive the classical equations of motions as@xcite@xmath110 _
{ n+1}=\left [ \begin{array}{c } z\cos \left ( \kappa x\right ) + y\sin \left ( \kappa x\right ) \\
-z\sin \left ( \kappa x\right ) + y\cos \left ( \kappa x\right ) \\
-x% \end{array}% \right ] _ { n}.\]]therefore , in the classical limit
, the dynamics of the qkt is governed by the above equation , and by using eq.([xyz ] ) , we plot the stroboscopic dynamics of the classical variables @xmath105 and @xmath106 in fig.[classical_map ] .
each point represents a state of @xmath111 in the phase space . in this plot , we choose @xmath112 , and thus there are periodic , quasi - periodic , and chaotic regions in the phase space .
the quantum evolution of the qkt is studied when @xmath15 is finite , and the chaos is indicated by the stroboscopic plot shown in fig.classical_map .
thus , to make connection between quantum and classical evolutions , the initial state should be chosen as coherent spin state ( css)@xcite , which can be viewed as a most classical state@xmath113where @xmath114 is the eigenstate of @xmath115 with the eigenvalue @xmath15 , and the rotation operator is defined as @xmath116 \right\ } , \]]where @xmath117 and @xmath118 .
the expectation values of the angular momentums on this state are given by @xmath119which is the same as eq.([xyz ] ) .
this is the reason that we choose css to be the initial state .
since css can be regarded as a classical state , and it is a product state , thus there is no correlations between qubits . substituting eq .
( [ collective ] ) into eq .
( [ eq37 ] ) , we can see that the operator @xmath120 can be written as a direct product of @xmath17 operators acting independently on each qubit . and note that @xmath114 is a product state , and thus the css is also a product state . in the classical limit
, the initial classical state is indeed a css , and during the evolution , the state is still classical .
however , in the quantum case , the factorization([jxy ] ) is not valid , since quantum correlations are created during the evolution , and thus the state is no longer a css .
such quantum correlations are characterized below by spin squeezing and concurrence .
, and @xmath121 with different initial states .
we choose @xmath122 in the periodic region ( a ) , @xmath123 @xmath124 and @xmath125 are almost the same , while @xmath121 is very large .
this implies that , the maximal spin squeezing is around the @xmath29-direction .
however , in the chaotic region ( b ) , @xmath126 the differences between @xmath127 and @xmath128 are obvious , that means , the maximal spin squeezing is not restricted to the @xmath29-direction . the dashed line corresponds to @xmath129 .
, width=302 ] at first , we use spin squeezing to characterize quantum correlations and quantum chaos . according to the discussions of sec.ii
, the spin squeezing parameter @xmath3 characterizes the minimal pairwise correlation , the parameter @xmath0 characterizes the minimal pairwise correlation in the plane which is perpendicular to the mean spin direction @xmath6 , and the parameter @xmath130 is the pairwise correlation along @xmath6-direction .
the numerical results of quantum evolutions of spin squeezing parameters @xmath3 , @xmath0 , and @xmath131 are given in fig.[spinsqu ] , and we find that the maximal spin squeezing directions are strongly influenced by quantum chaos . here ,
we analyze these three spin squeezing parameters in the periodic region , as shown in fig.[spinsqu](a ) .
it can be seen that the spin squeezing parameter @xmath0 is much smaller than @xmath132 , but there are very slight differences between @xmath0 and @xmath21 , thus the maximal spin squeezing , which refer to the minimal pairwise correlation , is around the @xmath29-direction . in the chaotic case , as shown in fig.[spinsqu ] ( b ) , we can find that the parameter @xmath3 is much smaller than both @xmath13 and @xmath130 at some time , that means the maximal spin squeezing is along neither the @xmath29-direction nor the @xmath133-direction .
therefore , the directions of maximal spin squeezing are around the @xmath29-direction in the periodic case , while they are not restricted to the @xmath29-direction in the chaotic case . denotes the mean spin direction , @xmath35 and @xmath36 are two orthogonal directions which are perpendicular to @xmath6 . in the left plot ,
the initial state is in the periodic region , with @xmath134 . and in the right plot , the initial state is in the chaotic region , with @xmath135 .
, width=340 ] the directions of the maximal spin squeezing are calculated in fig.[spq_dir ] .
the @xmath6-axis in fig.spq_dir is the mean spin direction@xcite .
it can be written in the spherical coordinate as @xmath136where @xmath137 and @xmath138 are the polar and azimuthal angles , respectively .
they are calculated as @xmath139where @xmath140 .
the other two axes in fig.[spq_dir ] are chosen as@xmath141from fig.[spq_dir ] , it can be easily found that the directions of the maximal spin squeezing are around the @xmath29-direction in the periodic case .
but in the chaotic case , the directions are not concentrated in a certain direction .
thus the directions of maximal spin squeezing are strongly influenced by quantum chaos . , @xmath142 and @xmath143 with different initial states , and we choose @xmath122 in ( a ) and ( b ) , the initial state is in the periodic region , at @xmath144 .
the values of @xmath125 and @xmath143 are almost the same except for the first two kicks , but there are small differences between @xmath127 and @xmath143 . in ( c ) and
( d ) , we choose @xmath145 , which is centered in the chaotic region .
there are small differences between @xmath125 and @xmath143 , but the differences between @xmath127 and @xmath143 are large .
the dashed line corresponds to @xmath129 .
, width=340 ] then , we use concurrence to character the quantum chaos , and give comparisons among the spin squeezing parameters @xmath0 , @xmath146 , and concurrence . from ref.@xcite , when @xmath147 , the relation between @xmath0 and @xmath76 is shown as eq.([sq_con ] ) for states with a fixed parity , and when @xmath148 , the relation does not hold . in the qkt model , there is not a simple relation between spin squeezing and concurrence as the states here are more general than those with a parity . in order to make a direct comparison between spin squeezing parameter and concurrence , here , we introduce a new quantity to describe the concurrence , the form
is@xmath149where @xmath150which can be got from the eq.([conc ] ) without the max function , and when @xmath151 , @xmath152 , namely the state is pairwise entangled .
the numerical results of the dynamical evolutions of @xmath153 , @xmath0 , and @xmath154 in the qkt model are given in fig.[concurrence_spinsqu ] .
in the periodic case , as shown in fig.[concurrence_spinsqu](a ) and ( b ) , we can see that , at the first two kicks , there are small differences between @xmath3 @xmath155 and @xmath156 .
after the third kick , the parameters @xmath3 and @xmath156 are nearly coincide , while there are also small differences between @xmath0 and @xmath154 .
we also note that when @xmath157 , both @xmath3 and @xmath0 are smaller than @xmath129 , so both the two spin squeezing parameters can well describe the pairwise entanglement . in the chaotic case , as shown in fig.[concurrence_spinsqu](c ) and ( d ) , we can find that there are very small differences between @xmath60 and @xmath154 , but the differences between @xmath0 and @xmath154 are large .
we also observe that when @xmath151 , @xmath158 , and vice versa .
it means the spin squeezing parameter @xmath60 may indicate the pairwise entanglement . at the second and third kicks , the parameter @xmath151 , @xmath159 ,
namely , the spin squeezing parameter @xmath0 is not a good quantity to describe the pairwise entanglement in this chaotic case .
, ( b ) the quasi - periodic case @xmath160 , and ( c ) the chaotic case @xmath161 . here
, we choose @xmath162.,width=340 ] recently , entanglement sudden death and sudden birth have received a lot of attentions . in this paper , we find that both esd and esb occur in the qkt model . here
we study the influences of quantum chaos on esd and esb .
we choose the initial states are in the periodic , quasi - periodic , and chaotic regions , and the dynamics of concurrence are shown in fig.esd-esb . since the initial state is a css , there is no entanglement at first , after the first kick , the pairwise entanglement ( @xmath84 ) is produced .
as time evolves , the concurrence decreases to zero , and remains for a period of time . as shown in fig.[esd - esb ] ( c ) , we can see that , when the initial state is in the chaotic region , there is no entanglement again . however , we observe esb when the initial states are in the periodic and quasi - periodic regions , as shown in fig.[esd - esb ] ( a ) and ( b )
. the whole length of the time intervals for zero entanglement in the periodic case is shorter than that in the quasi - periodic case .
, ( b ) the quasi - periodic case @xmath163 and ( c ) the chaotic case @xmath145 . here
, we choose @xmath162.,width=340 ] similar to concurrence , we also find sssd and sssb in the qkt model . in ref.@xcite , the authors introduced a quantity to describe the spin squeezing , the form is @xmath164 therefore , if there is no spin squeezing , i.e. @xmath165 we have @xmath166 .
the numerical results of @xmath167 in the periodic , quasi - periodic , and chaotic cases are illustrated in fig.sssd-sssb . at first
, we consider the periodic case , as shown in fig.[sssd - sssb ] ( a ) , there is spin squeezing ( @xmath168 ) at the first kick , and it quickly decreases to zero . as time evolves , @xmath169 becomes larger than @xmath170 .
both sssd and sssb appear multiple times under this condition . in the quasi - periodic case , as shown in fig.[sssd - sssb](b ) , both sssd and sssb appear alternatively , but the whole length of time intervals for zero spin squeezing parameter is longer than that in the periodic case . at last ,
we consider the chaotic case , as shown in fig.[sssd - sssb](c ) , the spin squeezing vanishes after a few kicks , and then there is no spin squeezing again . from the above discussions , we find that the quantum chaos greatly affects the dynamics of spin squeezing and entanglement . when the initial states are in the periodic and quasi - periodic regions , both esd ( sssd ) and esb ( sssb ) appear alternatively , and when the initial states are in the chaotic region , there is only esd ( sssd ) .
in summary , we first prove that negative correlations are equivalent to spin squeezing for systems with only symmetric dicke states populated
. then we study the effects of quantum chaos on spin squeezing and entanglement in the qkt model . using the spin squeezing parameter @xmath171
, we find that , in the periodic case the directions of the maximal spin squeezing are around the @xmath172-direction , while in the chaotic case , they deviate from the @xmath173-direction and behave irregularly .
then , we study the dynamics of spin squeezing parameters @xmath0 , @xmath3 , and concurrence , and find that @xmath3 is a good quantity to characterize the pairwise entanglement .
at last , we study the influences of quantum chaos on esd ( sssd ) and esb ( sssb ) in the qkt model . we find that both esd ( sssd ) and esb ( sssb ) occur alternatively in the periodic and quasi - periodic cases , but there is only esd ( sssd ) in the chaotic case .
we believe that the behaviors of spin squeezing in the qkt model may also be found in other models that exhibit quantum chaos , e.g. the dicke model .
this work is supported by nsfc with grant no.10874151 , 10935010 , 10947019 ; and the fundamental research funds for the central universities .
a. s. srensen , l .- m .
duan , j. i. cirac , and p. zoller , nature ( london ) * 409 * , 63 ( 2001 ) ; a. s. srensen , phys .
a * 65 * , 043610 ( 2002 ) ; a. s. srensen and k. mlmer , phys .
lett . * 86 * , 4431 ( 2001 ) .
k. n. alekseev and d. s. primak , jetp * 86 * , 61 ( 1998 ) ; k. n. alekseev , opt
. commun . * 116 * , 468 ( 1995 ) ; k. n. alekseev and j. perina , phys .
e * 57 * , 4023 ( 1998 ) ; k. n. alekseev and j. perina , phys .
a * 231 * , 373 ( 1997 ) . c. thiel , j. von zanthier , t. bastin , e. solano , and g. s. agarwal , phys .
* 99 * , 193602 ( 2007 ) ; c. e. lpez , j. c. retamal , and e. solano , phys . rev .
a * 76 * , 033413 ( 2007 ) ; r. prevedel , g. cronenberg , m. s. tame , m. paternostro , p. walther , m. s. kim and a. zeilinger , phys .
lett . * 103 * , 020503 ( 2009 ) ; w. wieczorek , r. krischek , n. kiesel , p. michelberger , g. toth and h. weinfurter , phys .
lett . * 103 * , 020504 ( 2009 ) ; s. campbell , m. s. tame , and m. paternostro , new j. phys .
* 11 * , 073039 ( 2009 ) . | we study spin squeezing , negative correlations , and concurrence in the quantum kicked top model .
we prove that the spin squeezing and negative correlations are equivalent for spin systems with only symmetric dicke states populated .
we numerically analyze spin squeezing parameters and concurrence in this model , and find that the maximal spin squeezing direction , which refers to the minimal pairwise correlation direction , is strongly influenced by quantum chaos .
entanglement ( spin squeezing ) sudden death and sudden birth occur alternatively for the periodic and quasi - periodic cases , while only entanglement ( spin squeezing ) sudden death is found for the chaotic case . |
dynamic positron emission tomography ( pet ) can be used to measure tracer kinetics in vivo , from which physiological parameters , such as tissue perfusion , ligand receptor binding potential , and metabolic rate can be determined using compartmental modelling techniques .
estimation of the kinetic parametric images can be extremely challenging , since the data are often very noisy .
most conventional methods either define a region of interest ( roi ) and estimate parameters based on the averages@xcite , or in a voxelwise fashion . the former requires the identification of roi which itself is difficult .
the latter fails to utilize information from nearby voxels , resulting in more noisy estimates .
estimations are typically carried out using a minimum least - squares approach @xcite or a basis function approach@xcite . given low signal - to - noise ratio ( snr ) , particularly in voxelwise estimations ,
some external constraints are often necessary to stabilize parameter estimation .
smoothness regularization can be used@xcite , constraining the parameters from nearby spatial locations to be more similar .
similarly , tikhonov regularization @xcite can be used to directly constrain parameter values to be within a certain range , so that estimates obtained are less sensitive to noise . to account for irregularities in the noise distribution
, mixture models can be fitted to each voxel @xcite . in this case , it is necessary to restrict the total number of mixture components to be small and employ regularization to constrain parameter estimates . assuming a gaussian error distribution , bayesian methods @xcite provide an alternative way of obtaining uncertainty estimates for the kinetic parameters , as well as model choice for the competing compartmental models . however , these methods yield higher voxel to voxel variability because each voxel was processed independently , and the assumption of gaussian distribution can also be inappropriate , leading to biased parameter estimates . this has led to the development of several approaches , in which a clustering method was performed first to cluster the pet images into several homogeneous regions , and kinetic parameter estimations were then performed afterward based on the averaged values of each cluster .
see , for example , hierarchical clustering of the time activity curves ( tacs ) using a weighted dissimilarity measure@xcite , and a comparison of a number of different hierarchical clustering algorithms in this context@xcite .
recently , simultaneous clustering and parameter estimation methods have been proposed using a spatially regularized @xmath1-means algorithm .
the algorithm iteratively estimates the kinetic parameters in a least squared sense between each cluster update@xcite .
it was demonstrated that incorporating the physiological model in the clustering procedure performed better than their counterparts in terms of clustering .
however , the method offers no guidance on the choice of cluster numbers , or how to select the spatial regularization parameter , both can have great influence on the results .
a similar algorithm was proposed @xcite , where the clustering and parameter estimation were performed simultaneously , although spatial correlation was ignored .
they demonstrated improvements to parameter estimation in myocardial perfusion pet imaging .
we develop a fully bayesian approach , based on defining a finite mixture of multivariate gaussian distributions to model each voxelwise tac .
we consider that there are a number of distinct homogeneous groups of voxel kinetics which tend to be spatially contiguous .
the optimal number of mixture components ( or groups ) is estimated via information theoretic criteria .
this provides a flexible specification of the error distribution for the tac .
additionally , we model the spatial dependence between the tacs via the markov random field ( mrf ) , which allows us to borrow information across nearby voxels .
our model simultaneously handles both spatial and temporal information , making full use of the data available , and this is done with the estimation of the kinetic parameters in a single step .
we apply our approach to simulated one - compartment pet perfusion data and compare the performance of our approach with both the standard voxelwise curve - fitting approach and the spatial temporal approach @xcite , using the true kinetic parameters as the gold standard .
we also apply our method to an in vivo pig study data .
all the simulation studies were performed using an ncat torso phantom @xcite which consists of heart , lungs , liver , and soft - tissue compartments . the left ventricle ( lv ) myocardium was segmented into 17 standard segments.@xcite the simulation was based on @xmath2f - flurpiridaz , which is a new myocardial perfusion tracer that exhibits rapid uptake and longer
washout in cardiomyocytes . based on the one - tissue compartmental model ,
the tac of the tissue concentration , @xmath3 , was simulated using @xmath4,\ ] ] where @xmath5 is the blood input function , @xmath0 and @xmath6 are kinetic rate constants for the segment , and @xmath7 denotes convolution operation .
the input function used in the simulation was based on a previously published @xmath2f - flurpiridaz study@xcite . during the study ,
the lv input function was extracted with generalized factor analysis on dynamic series@xcite .
this lv input function was treated as the plasma input function .
the kinetic parameters , i.e. , @xmath8 , assigned to 17 segments were based on the realistic values obtained from pet perfusion studies on normal patients .
@xcite in order to mimic a myocardial defect , the segment located in the anterior wall was assigned with values by lowering @xmath0 and @xmath6 by 50% and 20% , respectively , of their original values .
we added the @xmath9 segment to include other voxels not part of the left ventricle myocardium .
table [ segmentname ] shows the kinetic parameters assigned to all the 18 segments in the myocardium .
the blood input function @xmath10 and tacs for one normal ( basal inferoseptal ) and one defect ( apex ) segments are shown in figure [ inputsample ] .
+ [ hbt ] .
segment names and their assigned @xmath0 values in ml / min / cc , @xmath6 values in 1/min ( i.e.,the ground truth ) . [ cols="<,^,^,<,^,^",options="header " , ] a system matrix corresponding to philips gemini pet - ct camera , which includes position dependent point spread function modelling , a forward - projection operator implemented using siddon s method , line of response ( lor ) normalization factors , and attenuation correction factors , was used to create noise - free sinograms from tacs .
@xcite the simulated sinogram data is equivalent to a 13-min dynamic pet scan with the framing scheme of 6 @xmath11 5s , 3 @xmath11 30s , 5 @xmath11 60s , and 3 @xmath11 120s frames .
twenty five dynamic pet noise realizations were generated .
both random and scatter events were not included in this study .
the total number of events simulated in all the time frames is 50 m. the decay of the tracer was not simulated .
poisson noise was then added to each pixel in the sinogram based on the mean counts for the pixel . for each noise realization
, the image reconstruction at each time frame was performed using standard ordered subset expectation maximization @xcite(osem ) with 16 subsets and 8 iterations .
no postreconstruction smoothing was applied .
the physical dimension in the image reconstruction was 57.6 cm @xmath11 57.6 cm @xmath11 16.2 cm , matrix dimension was 128 @xmath11 128 @xmath11 36 , where the voxel size was 0.45 cm @xmath11 0.45 cm @xmath11 0.45 cm .
a pig with a body weight of 40 kg was scanned on a siemens biograph truepoint pet / ct with the radiotracer @xmath2f - flurpiridaz .
first , a planar x - ray topogram was performed to allow delineation of the field of view ( fov ) and centering on the heart following ct and pet acquisitions .
the cardiac ct was used for structure localization and later for attenuation correction during reconstruction of pet images .
emission pet data were acquired in 3d list mode and started concomitantly to the injection of @xmath2f - flurpiridaz , the injected activity was 11 mci at the time of injection .
list mode data were framed into dynamic series of 12 x 5 , 8 x 15 , 4 x 30 , 5 x 60s .
pet images were reconstructed using filtered back projection with minimal filtering ( voxel size : 2.14x2.14x3 mm3 , 55 slices ) .
attenuation correction was obtained from the ct images .
decay correction was applied and the first 10 min of the data are used for kinetic analysis .
the input functions for the left and right ventricle were obtained by averaging the tacs from a manually defined region . a one - compartment model with spill - over correction was used .
the described experiment was performed under a protocol approved by the institutional animal care and use committee at the massachusetts general hospital .
kinetic analysis is performed by curve - fitting the tac in each voxel using a nonlinear least - square fitting , @xmath12 where @xmath13 is the reconstructed activity concentration for voxel @xmath14 at time frame @xmath15 divided by frame duration @xmath16 , @xmath17 } ds$ ] , is the average concentration over time frame @xmath15 using the current estimates of the kinetic parameters @xmath18 in voxel @xmath14 , and measured blood input function @xmath19 , @xmath20 is the weighting factor which herein is chosen to be the squared frame duration divided by the total counts in that frame @xcite .
this nonlinear least - square problem can be solved using the levenberg - marquardt algorithm .
@xcite we denote this standard curve - fitting ( scf ) approach in this paper .
the spatial skms method performs spatial @xmath1-means clustering and parameter estimation iteratively @xcite .
the process is as follows .
( 1 ) initialize the cluster means @xmath21 for a predetermined number of clusters @xmath22 . ( 2 ) for each @xmath23 , estimate kinetic parameters @xmath24 argmin@xmath25 ; subject to positivity constraints on @xmath26 . ( 3 ) for each voxel indexed by @xmath27 , reassign cluster membership by minimizing the objective function @xmath28 .
@xmath29 is the tac at voxel @xmath14 .
@xmath30 determines the influence of the spatial regularizer .
@xmath31 is the indicator function returning a one if @xmath29 and @xmath32 belong to the same cluster and zero otherwise , for all @xmath32 in the neighbouhood of @xmath29 .
( 4 ) based on the new clusters , calculate @xmath33 as the mean for each cluster .
( 5 ) repeat above steps until there are no significant changes in @xmath34 .
there are two main issues for skms .
first , the authors offered no theoretical guarantee of convergence of their proposed algorithm .
second , both the number of clusters and spatial regularization parameter @xmath30 , need to be determined but it is not clear how this can be done . in our implementation of their method , we chose these parameters by looking at a range of @xmath30 and @xmath22 values , and selected the values which minimizes the errors with respect to @xmath0 parameter estimates , setting @xmath30 to 0.2 and @xmath22 to 17 .
we note , however , this procedure produces the best possible outcome for skms but is only possible for simulation data where we know the ground truth .
this method was not implemented for the pig study data . here , we describe our proposed modelling and estimation approach .
we denote the reconstructed activity concentration data by @xmath35 , for voxel @xmath14 .
each data point @xmath36 corresponds to the reconstructed activity concentration at time @xmath15 .
we assume that the data @xmath29 can be grouped into @xmath22 spatially homogeneous groups , where within each group , all voxels share the same kinetic rate parameters ( or tacs ) and their variations are only due to noise .
the number of groups , @xmath22 , is treated as unknown and is chosen by the information theoretic model selection criterion , bayesian information criterion ( bic ) .
we use a mixture of multivariate gaussian distribution with @xmath22 components to model the noisy data . given the roi used for the analysis may include some voxels outside the myocardium , as well as some noisy voxels with very little activity uptake inside the myocardium , we allow one component to cluster these types of voxels .
we call this the noise component .
the potts model @xcite is used to account for spatial correlation between the tacs .
this is achieved by introducing a set of auxiliary random variables @xmath37 , where @xmath38 takes one of the values @xmath39 , and represents the group / cluster membership for each voxel .
mathematically , each noisy tac is given by the mixture of @xmath40-dimensional gaussian , + @xmath41 where @xmath42 is the marginal density of the potts model , @xmath43 is the density of the multivariate gaussian , and @xmath30 is the parameter that reflects the spatial strength between voxels .
a value of 0 indicates independence between voxels , while larger values of @xmath30 will tend to cluster all voxels into one cluster .
the mean vector of the @xmath23th multivariate gaussian component is denoted by @xmath44 , and @xmath45 is the covariance matrix , assumed to be the same for all mixture components .
one may relax this assumption to allow more general covariance structure ; however , in our simulations studies , the same covariance structure worked well . here
, @xmath46 denotes the variance at time @xmath15 , and the data are assumed to be temporally independent as data were based on the reconstructed image at each time frame .
the mixture model representation allows the error distribution to be more flexible .
we refer to our model as the smm .
we set the mean vector for the noise component @xmath47 as @xmath48 where the @xmath49 are unknown parameters .
this component is dominated by noise , taking small values compared with other voxels with larger tac measurements . for the remaining components
@xmath50 , we model the mean vector as a function of the solution to the ordinary differential equation ( ode ) describing the one - compartment model @xcite , although extensions to more compartments are straightforward . hence ,
for @xmath51 , we set @xmath52 ds},\ ] ] where @xmath53 is a measured blood input function and @xmath54 is the duration of the @xmath15th time frame .
the parameters @xmath55 and @xmath56 are the kinetic rate parameters specific for group @xmath23 . for the pig study data analyses , we modified equation [ eqn : kinetic ] to account for spill - over effects , @xmath57 ds},\ ] ] where @xmath58 and @xmath59 denote the component specific spill - over fractions for the left and right ventricle respectively . @xmath60 and
@xmath61 were obtained by manually averaging the tacs from the appropriate rois .
@xmath53 was taken as @xmath60 multiplied by the plasma fraction , where the plasma concentration ratio was estimated based on blood samples drawn from previous studies . for bayesian inference , we need to specify prior distributions for the unknown parameters @xmath62 , @xmath63 , @xmath64 .
we assume independent and uninformative priors for all the parameters , so that the priors are broadly noninformative . for the kinetic rate parameters , we use the uniform distribution for all @xmath23 , @xmath65 and @xmath66 , where @xmath67 denotes uniform distribution .
we have used @xmath68 and @xmath69 in our simulation studies . in real applications
, one can sometimes get very abnormal rate constants , and a lower value of @xmath70 , such as 0.1 used for our pig study data , might be appropriate . setting @xmath70 much lower than the plausible ranges for @xmath0 will result in additional clusters of the noise voxels being estimated with the kinetic model , and will unnecessarily add to computational cost .
for the mean vector of the noise component @xmath71 , @xmath72 . setting the prior for @xmath0 sufficiently away from zero allows us to distinguish between the noise component and the non - noise components .
we set prior @xmath73 , where we take @xmath74 to be 1 , so as to include most of the plausible values of @xmath30 . finally , for the variance parameters @xmath46 , we follow the standard approach and use the usual vague conjugate prior with inverse gamma distribution @xmath75 , where @xmath76 for an uninformative prior on @xmath77 . for the pig study data ,
we define independent priors for the additional parameters @xmath78 and @xmath79 , and set @xmath80 and @xmath81 .
bayesian inference proceeds via the posterior distribution , obtained by the simple product of the likelihood and the priors in section [ sec : priors ] .
the likelihood function is given by @xmath82 the posterior distribution is given by the bayes theorem as the product of the likelihood and the priors @xmath83 where the term @xmath84 denotes the prior distribution .
mcmc algorithms were developed to sample from the posterior distribution , using a combination of random - walk metropolis - hastings and gibbs updates .
details for the implementation of the algorithm for the model in equation [ eqn : kinetic ] are given in the appendix , the model of equation [ eqn : kineticsp ] is a straight forward extension .
note that occasionally identifiability issues arise in the mcmc estimation of mixtures .
parameters from different components can switch labelling as a result of the invariance of the posterior distribution with respect to labelling .
this is not an issue when only the map estimates are required .
the simplest way to handle this is by imposing certain ordering constraints on parameters @xcite , or via postprocessing of the mcmc output@xcite . in this article
, the large number of mixture components was adequately handled using an efficient postprocessing algorithm for mcmc output @xcite .
one of the uncertainties of the above model is the selection of the value of @xmath22 , which plays a crucial role in the resulting parameter estimation . for model - based inference ,
in which a likelihood is readily available , a number of model selection criteria are available , including bayesian information criterion ( bic ) , integrated completed likelihood ( icl ) , deviance information criterion ( dic ) and akaike information criterion ( aic ) .
the bic is often considered to be more parsimonious and is the frequently adopted measure of goodness of fit of the model.@xcite our approach uses bic as the criterion to determine the optimal value of @xmath22 .
the bic @xcite is given as @xmath85 where @xmath86 is the likelihood function corresponding to the model with @xmath22 components , evaluated at the map estimator of @xmath87 and @xmath88 , the vector of all remaining unknown parameters .
@xmath89 is the number of parameters to be estimated , which includes all the unknown kinetic parameters for each cluster , the variance parameters , and any hyperpriors which are estimated .
@xmath90 is the number of observations or voxels .
bic penalizes models with too many parameters against the maximized log - likelihood ( or fit to the data ) .
optimal choice of @xmath22 corresponds to the model with the smallest bic value .
we tuned the gaussian random - walk proposal distributions to obtain an optimal overall acceptance probability of around 20%-40% . for the simulation data sets
, we used @xmath91 , @xmath92 , @xmath93 and @xmath94 .
for the pig data , we used @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 and @xmath100 for a full bayesian analysis of a single simulated data set , we ran mcmc for 10000 iterations with the first 4000 iterations discarded as burn - in , and we keep every tenth sample due to high autocorrelation in the mcmc sample . note that for map estimates , taken as the set of parameter values which gave the highest posterior probability during the mcmc run , we used only 6000 iterations , as mcmc chains can be expected reach modal regions of the posteriors quite quickly . for the real data set , 8000 iterations of mcmc were obtained with the first 6000 discarded as burn - in .
we first determine the number of components @xmath22 , by running mcmc for @xmath101 and computing the model selection criteria based on bic ( see section [ sec : g ] ) .
then for a fixed @xmath22 , we run posterior inference for a given dataset . for the evaluation of the proposed algorithm
, we used 25 replicate simulations . for each replication at the chosen value of @xmath22
, we obtain map estimators for comparison with scf and skms .
the performance of map is known to be worse than the posterior mean estimate , but it is sufficient to provide a good guide on the quality of the inference .
the results are presented in section [ sec : results ] . for a given noise realization , @xmath90 ,
we compute kinetic parameter bias of voxel @xmath14 using @xmath102 where @xmath103 is the bias of estimated kinetic parameter using the true kinetic parameter @xmath104 as gold standard .
based on 25 noise realizations , we compute the mean bias @xmath105 , the mean squared bias @xmath106 , and standard deviation @xmath107 bias for voxel @xmath14 using @xmath108 where @xmath109 is the total number of noise realizations .
we perform the calculations described above for scf , smm and skms methods and make a comparison between methods .
figure [ bicplot ] shows the bic values ( left panel ) and the corresponding log likelihood ( right panel ) for competing models for a single noise realization .
horizontal lines in both subfigures denote the minimum and maximum values for bic and the log likelihood respectively .
the log likelihood values are expected to keep increasing with @xmath22 , while the bic penalizes the use of additional parameters in models with larger @xmath22 .
both the bic and log likelihood changed dramatically from @xmath110 to about @xmath111 , preferring models with larger @xmath22 values , and this stabilized after around @xmath112 .
part of the changes seen here can be attributed to monte carlo errors .
thus , a parsimonious choice for @xmath22 would be @xmath112 , representing the model with 16 tac components and one noise component .
subsequent results for simulated data in this paper were generated by the model with @xmath112 .
the same value of @xmath22 was also found for the pig study data . for the simulation data , figures
[ k1plot_g17 ] and [ k2plot_g17 ] show the corresponding marginal posterior distributions of @xmath0 and @xmath6 respectively , with vertical lines indicating the posterior mean , and the uncertainty of the estimates indicated by the spread of the distributions . for @xmath113 clusters .
vertical dashed line denotes corresponding posterior means .
based on a single noise realization of simulation data .
, width=529,height=680 ] for @xmath113 clusters .
vertical dashed line denotes corresponding posterior means .
based on a single noise realization of simulation data.,width=529,height=680 ] to assess the robustness of our estimation procedure and its performance against existing methods , we repeated our estimation procedure for 25 replicate data sets , obtained from the same simulation setup .
we implemented the three competing methods smm , scf and skms .
there was a single extremely large value of @xmath6 estimate from skms , which we omit from the results shown .
figure [ kbias ] shows the distribution of the mean squared biases of @xmath0 and @xmath6 in the abnormal , normal and noise regions , and the overall standard deviation of the biases .
the computations were calculated according to equation [ eqn : tbs ] , with the exception that in the noise region , the bias was computed by setting the denominator of equation [ eqn : bias ] , @xmath114 , to 1 , since we can not divide by zero .
results for the abnormal roi are shown in the first row of figure [ kbias ] . for @xmath0 ,
the mean squared biases for smm ranged from 0.006 to 0.12 , with a median of 0.04 .
for scf , the range was from 0.013 to 0.175 , with a median of 0.06 .
for skms , the range was between 0.02 and 0.39 , and the median was 0.06 . for the @xmath6 estimation , the biases ranged from 0.06 to 1.12 for smm , the median was 0.25 .
for scf , the range was between 0.13 and 1.47 , and the median was 0.4 . for skms , the biases ranged from 0.14 to 0.7 , and the median was 0.31 .
the normal roi is shown in the second row of figure [ kbias ] . for @xmath0 , the mean squared biases for smm ranged from 0.006 to 0.29 , with a median of 0.03 . for scf ,
the range was 0.005 to 0.27 , with a median of 0.03 . for skms ,
the range was between 0.007 and 0.33 , and the median was 0.04 .
for the @xmath6 estimation , the biases ranged from 0.01 to 0.37 for smm , with a median of 0.09 . for scf , the range was between 0.01 to 055 , and a median of 0.1 .
for skms , the biases ranged from 0.02 to 0.32 , and the median was 0.09 . the noise region is shown in the third row of figure [ kbias ] . for @xmath0 ,
the mean squared biases for smm ranged from 0.004 to 0.165 , with a median of 0.007 .
for scf , the range was 0.0001 to 3.32 , with a median of 0.02 .
for skms , the range was between 0.0005 and 0.22 , and a median of 0.05 . for the @xmath6 estimation ,
the biases were approximately 0 for smm .
for scf , the biases ranged between 0 to 0.06 , and the median was 0.03 . for skms , the biases ranged from 0 to 13.05 , and the median was 2.96 .
the last row of figure [ kbias ] shows the standard deviations of the biases for @xmath0 and @xmath6 . for @xmath0 , the standard deviations of biases ranged from 0 to 0.34 for smm , with a median of 0 . for scf ,
the range was between 0.008 to 1.70 , with a median of 0.13 .
for skms , the range was between 0.073 to 0.63 , and the median was 0.16 . for the @xmath6 standard deviations of bias , the range was between 0 and 0.93 for smm , with a median of 0 .
for scf , the range was between 0.002 to 1.19 , and a median of 0.01 .
for skms , the range was between 0.02 and 2.89 , with a median of 1.20 .
figure [ petereu ] compares the bias and standard deviation of bias between smm , scf and skms for a single slice .
the bias is calculated according to the first term in equation [ eqn : tbs ] , this is the average of the biases over 25 replications .
figure [ k1graph ] shows the @xmath0 estimates .
the biases ranged from -0.27 to 0.10 , -0.28 to 0.09 and -0.33 to 0.08 respectively , for smm , scf and skms .
similarly , the standard deviations ranged from 0 to 0.26 , 0 to 0.37 and 0 to 0.35 respectively . for the @xmath6 estimates in figure [ k2graph ] ,
the biases ranged from -0.22 to 0.2 , -1 to 1.06 and -0.46 to 0.28 respectively .
the standard deviations ranged from 0 to 0.59 , 0 to 0.86 and 0 to 2.89 respectively .
figure [ petereupig ] compares a single slice of the kinetic parametric images between smm and scf for the pig study .
@xmath0 parameters were constrained to be between @xmath115 and @xmath116 in both smm and scf estimation , as unconstrained estimation lead to many physiologically implausible large values of @xmath0 . and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] + and @xmath6 in the abnormal , normal , and the noise rois respectively . the last row shows the standard deviation of the biases
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] + and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] + and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] and @xmath6 in the abnormal , normal , and the noise rois respectively .
the last row shows the standard deviation of the biases .
mean squared biases and the standard deviation of biases are calculated according to equations [ eqn : bias ] and [ eqn : tbs ] , over 25 replicate simulation data sets.,title="fig : " ] 0.45 0.45
this paper proposes a novel method , smm , that clusters voxelwise tacs and estimates kinetic parameters simultaneously .
our modelling approach shares similarities to the recently proposed work @xcite , where the mixture model was fitted to each voxel ( while still borrowing information across nearby voxels ) to overcome the issue of non - gaussian error distributions .
we allow several similar voxels to share the same parameter values , since separate mixture models fitted to each voxel introduces too many parameters , and will lead to more estimation uncertainty .
our approach naturally allows us to constrain parameter estimates without the need to specify regularization parameters as in the usual bayesian maximum a posterior ( map ) approaches .
finally , we allow the data to determine the most appropriate number of mixtures to fit to the data .
our model - based approach offers several advantages , compared to other existing statistical approaches described above .
we require minimal user input in the algorithm , preferring to allow the data to dictate the optimal choices .
one benefit of our modelling approach is in the determination of the optimal number of mixture components .
we also automatically compute the value of the smoothing parameter used in the mrf model .
this unknown parameter is difficult to estimate , and in many applications of spatial modelling , the estimation of this parameter has not been carefully considered .
the choice of both these parameters can have a big impact on results , since suboptimal choices will either result in higher bias or higher variance for the resulting parameter estimates .
finally , the bayesian statistical framework allows us to quantify uncertainty probabilistically , since uncertainty in the model and parameters is a natural consequence of our modelling framework .
an efficient mcmc algorithm allows us to provide parameter estimates , as well as uncertainty quantification simultaneously . in figure
[ bicplot ] , the bic values start to reach a minimum at around @xmath11716 or 17 .
we chose to work with 17 , but higher values of 18 or 19 will work equally well .
these numbers are similar to the number of true segments simulated ; however , we expect that this number can be different depending on the nature of the noise
. figures [ k1plot_g17 ] and [ k2plot_g17 ] indicate the component mean estimates for @xmath0 and @xmath6 .
it is difficult to make direct correspondences between the clusters we obtained with the true segments .
the first four components correspond mostly to noise , the next four components correspond mostly to abnormal voxels and the rest belong to normal voxels .
the discrepancy between the estimated values of @xmath0 and the truth is most obvious in the abnormal region , this is possibly a combination of partial volume effect , as well as misclassification of the normal voxels . given that smm outperforms the other two methods in the abnormal region , we believe similar issues with the data are affecting the other two methods also .
despite the fact that it is difficult to make sense of individual clusters , aggregating the clusters can provide us with information about the larger rois .
for instance , if we are interested in identifying the three regions of noise , abnormal and normal , we can aggregate the clusters according to @xmath118 , @xmath119 and @xmath120 respectively .
a similar procedure can be used to classify the regions using the results from scf and skms . in terms of misclassification rates ,
based on a single simulation data set , smm classified 96.34% of noise voxels correctly , compared to 94.43% and 87.60% for scf and skms .
the misclassified voxels for smm were all assigned to the abnormal voxels , this corresponds to the first four clusters in figure [ k1plot_g17 ] . for the other two methods
, they were spread between abnormal and normal voxels . for the abnormal region ,
smm had a 100% correct classification , while this was only 62.68% for scf and 52.82% for skms .
all of the misclassifications in scf and skms were allocated to noise .
finally , for the normal region , smm , scf and skms had 69.57% , 73.49% and 58.99% respectively for correct classifications , most of the misclassifications were found to be allocated to the abnormal region . in terms of @xmath6 estimation
, smm is clearly better than the other two methods .
this can be seen clearly in figures [ kbias ] and [ petereu ] .
skms performed the worst , particularly in the abnormal and noise regions , their parameter estimation can be prone to very large biases . in the noise region , in particular , the median mean squared bias was around 2.96 , while the other two methods were close to 0 . putting constraints on these parameters may prove useful . for @xmath0 estimates ,
skms was marginally worse than the other two methods in terms of mean squared bias . in the abnormal region , smm shows noticeably superior performance , where it can be seen in figure [ kbias ] , top row , the entire distribution of smm is closer toward 0 than the other two methods .
the difference in the normal region is less obvious .
the third row in figure [ kbias ] shows the biases in the noise region ; here , since smm set @xmath0 in this region to 0 , the graph can be interpreted by looking separately at the values of mean squared bias below @xmath121 and above . on average , voxels with bias greater than this value
are essentially misclassified , i.e. , they should be singled out as noise , but instead have significant values for the kinetic parameters . for smm
, there was an average misclassification rate of 3.66% , for scf it is 5.71% and 11.65% for skms .
scf has the largest mean squared biases here , going up to 3.32 , while the other two methods remain around 0.2 . in terms of the standard deviations of the bias
, smm performed the best , while scf was the worst .
the plot in the last row of figure [ kbias ] shows that for @xmath0 , the range for scf goes up to 1.7 , while for smm and skms , this was only 0.34 and 0.63 respectively .
in fact , 15% of the voxels estimated by scf was greater than 0.34 ( the largest value obtained by smm ) , and 0.8% from skms .
the proposed method is clearly superior in terms of robustness , indicated by the substantially smaller standard deviation estimates , as can be seen in both figure [ kbias ] and figure [ petereu ] .
it also performed at least as good as , and sometimes better than the other two methods in terms of mean squared bias . in the single slice plot in figure [ petereu ] , where the mean of the raw biases was plotted , it is difficult to distinguish between the three methods .
this is due to the fact that when raw biases are averaged , they will go toward zero as the effects of the large positive and negative biases cancel out .
this will be true for all unbiased estimators regardless of how sensitive the estimations are to noise . in this sense ,
it is more useful to look at the mean squared or absolute biases . in terms of the estimation of the pig study data , it was not possible to compare the bias and standard deviations of the biases because the ground truth was not known . however , the parametric images shown in figure [ petereupig ] suggest that much smoother @xmath0 and @xmath6 images were produced by smm compared to scf .
the white color in the @xmath0 images indicates a value close to 1 , which is the upper bound of the artificial constraint we used .
it is clear from the figure that many values produced by the scf method were simply truncated at this value .
smm estimation produced significantly less values close to the upper bound .
the upper bound of 1 for @xmath0 is essentially arbitrary .
for smm estimation , if we remove this bound , we obtain two groups of voxels with physiologically implausibly high @xmath0 values whilst the rest of the voxel estimates remains unchanged , well below 1 .
however , in terms of the scf estimation , raising the bounds to higher than 1 produced many more voxels between 1 and 2 .
however , since this was a resting pig , where the mean blood flow at rest is around 0.65 ml / min / cc , we do not expect flow to be above 1 at rest , so the scf results with higher bounds would be difficult to interpret , since the higher values could also be due to spill - over from blood - pools , or voxels actually containing blood or noise . in terms of computation , it took about 4 hours to complete all 6000 iterations for each noise realization of simulated data , using matlab r2014b , running on a single node of the linux computational cluster katana at unsw , australia .
this is equivalent to running on an average pc .
the total number of voxels was 5746 .
we found that all the parameters converged quite quickly .
for scf and skms , the computational time was around 1 minute .
for the pig study data involving 16821 voxels , and a longer time series involving 29 time points , the computational times were 23.5 hours and 6 hours for smm and scf respectively .
we note that although smm is computationally more expensive , it provides additional uncertainty estimation , which the other two faster methods do not .
parallel computation or other computational methods , such as variational bayes @xcite , can be adopted to further speed up this process . in the future
, there are three directions to be considered to further develop our approach .
first , we can relax the within cluster homogeneity assumption .
this is easily achievable by relaxing the mean of the normal mixture to allow them to vary for each voxel observation .
however , this substantially increases the number of parameters that needs to be estimated and presents a computational challenge .
second , we can consider the use of sinogram data rather than reconstructed data to estimate kinetic parameters , and this can reduce the additional noise introduced through the reconstruction step ; however , this approach can be computationally challenging for full bayesian analyses .
third , given its flexibility , smm can be easily extended to more advanced kinetic models , such as the two - compartment tissue model without too much modification .
although we assess performance using simulations of cardiac perfusion pet imaging and demonstrate in vivo data for this application , our approach is not limited to this specific context and may also benefit other dynamic pet procedures as well as dynamic spect , dynamic contrast enhanced ct ( dce - ct ) , and dynamic contrast enhanced mr ( dce - mr ) .
this paper proposed a novel spatiotemporal approach , smm , to infer parametric pet images . by borrowing information from nearby voxels , smm can be used to simultaneously estimate kinetic parameters and classify voxels with similar kinetic parameters into spatially homogeneous groups .
we adopted the mrf to incorporate the spatial dependence of voxels .
we developed an efficient mcmc algorithm for the computation , which estimates all unknown parameters , including the notoriously difficult spatial smoothness parameter @xmath30 in the potts model .
the method provides parameter uncertainty estimation , as well as a principled way to determine the optimal number of voxel groups .
we used simulated cardiac perfusion pet data to evaluate the performance of smm and compared them with scf and skms .
smm was substantially less sensitive to noise than the other methods , it also yielded an overall smaller bias than scf and skms . in the pig study data
, smm was shown to produce smoother parametric images compared to the standard curve fitting .
although simulation and experimental data were based on cardiac pet studies of a one - compartment model , the approach may benefit other dynamic pet procedures , as well as more complex compartmental models .
this research was supported in part by unsw 2014 science silver star grant and nih grants r01-hl118261 , r01-hl110241 , and nih t32eb013180 .
we use mcmc for sampling from the joint posterior distribution of @xmath122 and all other parameters , given by equation [ eqn : post ] .
the prior distribution @xmath84 is taken as product of the individual prior components @xmath123 , as defined in section [ sec : priors ] . the first term on the right side of equation [ eqn : post ] is given by @xmath124 and the second term on the right side of the equation is given by @xmath125 this is the potts model , where @xmath37
, @xmath31 denotes indicator function taking value 1 if @xmath126 and 0 otherwise , and @xmath127 denotes the voxels @xmath128 in the neighbourhood of voxel @xmath14 . the partition function @xmath129 is estimated offline using thermal dynamic integration .
@xcite we use an 8 nearest neighbour first order structure for the potts model . + our computational algorithm for the one - compartmental model in equation
[ eqn : kinetic ] proceeds as follows : * step 2 * update @xmath132 , for @xmath50 . simulate a new value @xmath133 and compute @xmath134 with @xmath135 , according to equation ( [ eqn : kinetic ] ) . set @xmath136 to @xmath137 with probability @xmath138 , where @xmath139 with @xmath140 .
29ifxundefined [ 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 * * ( ) in @noop _ _ ( , ) pp . `` , '' ( , ) chap . , pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ ( , ) pp . @noop * * , ( ) _ _ , @noop ph.d .
thesis , ( ) @noop * * , ( ) @noop ( ) in @noop _ _ ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) | * purpose : * estimation of parametric maps is challenging for kinetic models in dynamic positron emission tomography .
since voxel kinetics tend to be spatially contiguous , the authors consider groups of homogeneous voxels together .
the authors propose a novel algorithm to identify the groups and estimate kinetic parameters simultaneously .
uncertainty estimates for kinetic parameters are also obtained .
+ * methods : * mixture models were used to fit the time activity curves . in order to borrow information from spatially nearby voxels ,
the potts model was adopted .
a spatial temporal model was built incorporating both spatial and temporal information in the data .
markov chain monte carlo was used to carry out parameter estimation .
evaluation and comparisons with existing methods were carried out on cardiac studies using both simulated data sets and a pig study data .
one - compartment kinetic modelling was used , in which @xmath0 is the parameter of interest , providing a measure of local perfusion . +
* results : * based on simulation experiments , the median standard deviation across all image voxels , of @xmath0 estimates were 0 , 0.13 , and 0.16 for the proposed spatial mixture models ( smms ) , standard curve fitting and spatial @xmath1-means methods respectively .
the corresponding median mean squared biases for @xmath0 were 0.04 , 0.06 and 0.06 for abnormal region of interest(roi ) ; 0.03 , 0.03 and 0.04 for normal roi ; and 0.007 , 0.02 and 0.05 for the noise region . +
* conclusions : * smm is a fully bayesian algorithm which determines the optimal number of homogeneous voxel groups , voxel group membership , parameter estimation and parameter uncertainty estimation simultaneously . the voxel membership can also be used for classification purposes . by borrowing information from spatially nearby voxels , smm
substantially reduces the variability of parameter estimates . in some rois ,
smm also reduces mean squared bias .
* keywords * pet image , kinetic model , myocardium , spatial mixture model , mcmc . |
current implementations of numerical homotopy algorithms @xcite such as phcpack @xcite , hom4ps @xcite , bertini @xcite , and nag4m2 @xcite routinely and reliably solve systems of polynomial equations with dozens of variables having thousands of solutions . here , ` solve ' means ` compute numerical approximations to solutions . ' in each of these software packages , the solutions are validated heuristically often by monitoring iterations of newton s method .
this works well in practice , giving solutions that are acceptable in most applications .
however , a well - known shortcoming of numerical methods for computing approximate solutions to systems of polynomials is that the output is not certified .
this restricts their use in some applications , including those in pure mathematics .
the program * alphacertified * is intended to remedy this shortcoming . in the 1980
s , smale @xcite and others investigated the convergence of newton s method , developing @xmath0-theory ( * ? ? ? * ch .
this refers to a computable positive constant @xmath1 depending upon a system @xmath2 of polynomials and a point @xmath3 such that , if @xmath4 then iterations of newton s method starting at @xmath5 will converge quadratically to a solution to @xmath6 , which is a point @xmath7 with @xmath8 . in principle
, smale s @xmath0-theory provides certificates for validating numerical computations with polynomials .
current implementations of numerical homotopy algorithms do not incorporate @xmath0-theory to certify their output or their path - tracking .
there have been two projects which use fixed double precision and focus on certified path - tracking .
malajovich @xcite released the most recent version of his polynomial system solver in 2003 , which uses @xmath0-theory to certify toric path - tracking algorithms , but he states that `` [ it ] is actually not intended for an end user . ''
beltrn and leykin @xcite have recently shown how to use @xmath0-theory to certify path - tracking , and hence the output of numerical homotopy algorithms .
while they demonstrate that certification can dramatically affect the speed of computation , this is an important development , as certified path - tracking is necessary for applications such as numerical irreducible decomposition @xcite or computing galois groups @xcite .
they are continuing this line of research .
we describe a program , * alphacertified * , that implements elements of @xmath0-theory to certify numerical solutions to systems of polynomial equations using both exact rational and arbitrary precision floating point arithmetic .
as it only certifies the output of a numerical computation , it avoids the bottlenecks of certified tracking , while delivering some of its benefits .
given a square polynomial system @xmath9 , * alphacertified * uses smale s @xmath0-theory to answer the following three questions for a finite set of points @xmath10 : 1 . from which points of @xmath11 will newton s method converge quadratically to some solution to @xmath6 ?
from which points of @xmath11 will newton s method converge quadratically to distinct solutions to @xmath6 ?
if @xmath6 is real ( @xmath12 ) , from which points of @xmath11 will newton s method converge quadratically to real solutions to @xmath6 ? often , a sharp upper bound @xmath13 on the number of roots to a square polynomial system @xmath6 is known .
given a set of @xmath13 points , * alphacertified * can be used to certify that iterations of newton s method starting from each point in the set converge quadratically to some solution to @xmath6 and that these solutions are distinct .
this guarantees that each of the @xmath13 roots of @xmath6 can be approximated to arbitrary accuracy using newton s method .
moreover , * alphacertified * can certify how many of the @xmath13 solutions to @xmath6 are real when @xmath6 is real . a polynomial system @xmath14 is _ overdetermined _ if @xmath15 , that is , if the number of polynomials exceeds the number of variables .
dedieu and shub @xcite studied newton s method for overdetermined polynomial systems and gave conditions which guarantee quadratic convergence of its iterations . unlike square systems , the fixed points of this overdetermined newton s method need not be solutions .
for example , @xmath16 is a fixed point of newton s method applied to @xmath17 $ ] .
the program * alphacertified * validates solutions to overdetermined systems .
given a finite set @xmath10 and an overdetermined system , it generates two or more random square subsystems , answers the three questions above for each , and compares the results . in particular , given @xmath18 , it can certify that , for a given approximate solution to two or more random subsystems , the associated solutions all lie within a distance @xmath19 of each other .
for a given @xmath19 , this heuristically validates solutions to overdetermined systems . in summary ,
* alphacertified * is novel in each of the following ways .
it implements algorithms from @xmath0-theory using either exact rational or arbitrary precision floating point arithmetic . when using exact rational arithmetic with a square polynomial system , its implementation of @xmath0-theory is completely rigorous .
it certifiably determines if an approximate solution corresponds to a real solution , which may be used to count the real solutions to a polynomial system , and it uses @xmath0-theory to obtain information on the roots of overdetermined systems .
the examples we give demonstrate the practicality of certification based on @xmath0-theory , and its viability as an alternative to exact symbolic methods , as the certificates for square systems when using exact rational arithmetic are mathematical proofs of computed results . in section [ sec :
alpha ] , we review the concepts of @xmath0-theory utilized by * alphacertified*. section [ sec : square ] presents the algorithms for square polynomial systems while section [ sec : overdet ] describes our approach to overdetermined polynomial systems .
implementation details are presented in section [ sec : details ] with examples presented in section [ sec : ex ] verifying some computational results in kinematics and generating evidence for conjectures in enumerative real algebraic geometry .
we summarize key points of smale s @xmath0-theory for square polynomial systems that are utilized by * alphacertified*. more details may be found in ( * ? ? ?
let @xmath9 be a system of @xmath20 polynomials in @xmath20 variables with common zeroes @xmath21 , and let @xmath22 be the jacobian matrix of the system @xmath6 at @xmath5 . consider the map @xmath23 defined by @xmath24 the point @xmath25 is called the _
newton iteration of @xmath6 starting at @xmath5_. for @xmath26 , let @xmath27 be the @xmath28 newton iteration of @xmath6 starting at @xmath5 .
[ defn : approxsoln ] let @xmath9 be a polynomial system .
a point @xmath3 is an _ approximate solution _ to @xmath6 with _ associated solution _
@xmath29 if , for every @xmath26 , @xmath30 that is , the sequence @xmath31 _ converges quadratically _ to @xmath32 . here , @xmath33 is the usual hermitian norm on @xmath34 , namely @xmath35 .
smale s @xmath0-theory describes conditions that imply a given point @xmath5 is an approximate solution to @xmath6 .
it is based on constants @xmath1 , @xmath36 , and @xmath37 .
if @xmath22 is invertible , these are @xmath38 if @xmath39 is such that @xmath22 is not invertible , then we define @xmath40 and @xmath41 . otherwise ,
if @xmath42 and @xmath22 is not invertible , then we define @xmath43 . in the formula for @xmath37 , the @xmath28 derivative @xmath44 ( * ? ? ?
5 ) to @xmath6 is the symmetric tensor whose components are the partial derivatives of @xmath6 of order @xmath45 .
it is a linear map from the @xmath45-fold symmetric power @xmath46 of @xmath34 to @xmath34 .
the norm in is the operator norm of @xmath47 , defined with respect to the norm on @xmath46 that is dual to the standard unitarily invariant norm on homogeneous polynomials @xcite , @xmath48 where @xmath49 is an exponent vector of non - negative integers with @xmath50 , @xmath51 , and @xmath52 is the multinomial coefficient .
the following version of theorem 2 from page 160 of @xcite provides a certificate that a point @xmath5 is an approximate solution to @xmath6 .
[ thm : alpha ] if @xmath9 is a polynomial system and @xmath3 with @xmath53 then @xmath5 is an approximate solution to @xmath6 .
additionally , @xmath54 where @xmath29 is the associated solution to @xmath5 .
if @xmath55 , then @xmath5 may not be an approximate solution to @xmath6 .
for example , for @xmath56 , if @xmath57 , then @xmath5 is not an approximate solution to @xmath6 yet @xmath58 . for a polynomial system @xmath9 and a point @xmath3
, we say that @xmath5 is a _
certified approximate solution _ to @xmath6 if holds .
theorem 4 and remark 6 of ( * ? ? ?
8) give a version of theorem [ thm : alpha ] that * alphacertified * uses to certify that two approximate solutions have the same associated solution .
[ thm : alpharobust ] let @xmath9 be a polynomial system , @xmath3 with @xmath59 and @xmath29 the associated solution to @xmath5 . if @xmath60 with @xmath61 then @xmath62 is an approximate solution to @xmath6 with associated solution @xmath32 .
the constant @xmath37 encoding the behavior of the higher order derivatives of @xmath6 at @xmath5 is difficult to compute , but it can be bounded above . for a polynomial @xmath63 of degree @xmath64 ,
say @xmath65 , define @xmath66 then @xmath33 is the standard unitarily invariant norm on the homogenization of @xmath67 . for a polynomial system @xmath9 ,
define @xmath68\,,\ ] ] and for a point @xmath3 , define @xmath69 let @xmath70 be the @xmath71 diagonal matrix with @xmath72 where @xmath73 is the degree of @xmath74 .
if @xmath22 is invertible , define @xmath75 the following version of proposition 3 from
i-3 of @xcite gives an upper bound for @xmath37 .
[ prop : boundgamma ] let @xmath9 be a polynomial system with @xmath76 and @xmath77 .
if @xmath3 such that @xmath22 is invertible , then @xmath78
let @xmath9 be a square polynomial system and @xmath79 be a set of points .
we describe the algorithms implemented in * alphacertified * which answer the three questions posed in the introduction .
these algorithms are stated for a polynomial system with complex coefficients , but are implemented for polynomial systems with coefficients in @xmath80 $ ] using both exact and arbitrary precision arithmetic . for each @xmath81 ,
* alphacertified * first checks if @xmath82 . if @xmath83 , then * alphacertified * determines if @xmath84 is invertible .
if it is , * alphacertified * computes @xmath85 and upper bounds for @xmath86 and @xmath87 using the following algorithm .
procedure @xmath88 : : input : : a square polynomial system @xmath9 and a point @xmath3 such that @xmath22 is invertible .
output : : @xmath89 , @xmath90 , and @xmath91 , where @xmath91 is the upper bound for @xmath37 given in proposition [ prop : boundgamma ] .
the next algorithm uses theorem [ thm : alpha ] to compute a subset @xmath92 of @xmath11 containing points that are certified approximate solutions to @xmath6 .
procedure @xmath93 : : input : : a square polynomial system @xmath9 and a set @xmath79 .
output : : a set @xmath94 of approximate solutions to @xmath6 .
begin : : -0.1 in + 1 .
initialize @xmath95 . 2 . for @xmath96 , if @xmath97 , set @xmath98 , otherwise , do the following if @xmath99 is invertible : 1
. set @xmath100 . 2 .
if @xmath101 , set @xmath98 .
return : : @xmath92 as * alphacertified * uses the upper bound for @xmath37 of proposition [ prop : boundgamma ] , it may fail to certify a legitimate approximate solution @xmath5 to @xmath6 . in that case
, a user may consider retrying after applying a few newton iterations to @xmath5 .
the software * alphacertified * does not invoke an automatic refinement to inputs that it does not certify .
this is because newton iterations may have unpredictable behavior ( attracting cycles , chaos ) when applied to points that are not in a basin of attraction .
however , * alphacertified * does provide the functionality for the user to do this refinement .
suppose that @xmath5 is an approximate solution to @xmath6 with associated solution @xmath32 such that @xmath102 is invertible .
since @xmath5 is an approximate solution , @xmath103 converges to zero .
since @xmath37 is the supremum of a finite number of continuous functions of @xmath5 , @xmath104 is bounded .
in particular , @xmath105 converges to zero . given approximate solutions @xmath106 and
@xmath107 to @xmath6 with associated solutions @xmath108 and @xmath109 , respectively , theorems [ thm : alpha ] and [ thm : alpharobust ] can be used to determine if @xmath108 and @xmath109 are equal .
in particular , if @xmath110 then @xmath111 by theorem [ thm : alpha ] .
if on the other hand we have @xmath112 for either @xmath113 or @xmath114 , then @xmath115 by theorem [ thm : alpharobust ] .
this justifies the following algorithm which determines if two approximate solutions correspond to distinct associated solutions .
procedure @xmath116 : : input : : a square polynomial system @xmath9 and approximate solutions @xmath106 and @xmath107 to @xmath6 with associated solutions @xmath108 and @xmath109 , respectively , such that @xmath117 and @xmath118 are invertible .
output : : a boolean @xmath119 that describes if @xmath120 .
begin : : do the following : + 1 . for @xmath121 , set @xmath122 . 2 . if @xmath123 , * return * true .
3 . if @xmath124 and @xmath125 , for either @xmath126 or @xmath127 , * return * false . 4 . [ item : updatedistinct ] for @xmath121 , update @xmath128 and
return to ( a ) .
this will halt , determining whether or not @xmath129 as @xmath130 decreases quadratically with @xmath45 , while @xmath131 is bounded . a polynomial system @xmath9 is _ real _ if @xmath12 .
in that case , solutions to @xmath132 are either real or occur in conjugate pairs . also , @xmath133 for @xmath3 so that @xmath134 is a real map .
theorems [ thm : alpha ] and [ thm : alpharobust ] can be used to determine if an approximate solution of @xmath6 is associated to a real solution .
let @xmath5 be an approximate solution to @xmath6 with associated solution @xmath32 .
we do not assume that @xmath5 is real , for numerical continuation solvers yield complex approximate solutions .
by assumption , @xmath135 is also an approximate solution to @xmath6 with associated solution @xmath136 .
if @xmath137 then @xmath138 by theorem [ thm : alpha ] since @xmath139 consider the natural projection map @xmath140 defined by @xmath141 since @xmath142 , @xmath32 is not real if @xmath143 we have both a local and a global approach to show that @xmath32 is real .
for the local approach , theorem [ thm : alpharobust ] implies that @xmath144 is also an approximate solution to @xmath6 with associated solution @xmath32 if @xmath145 since @xmath146 is a real map and @xmath147 , this implies that @xmath148 .
we could also have showed that both @xmath5 and @xmath135 correspond to the same solution to deduce that @xmath149 . if @xmath150 then theorem [ thm : alpharobust ] implies that @xmath149 .
this is more restrictive then since @xmath142 . when @xmath151 , and yield closely related statements .
since @xmath152 we know that @xmath32 is real if @xmath153 and not real if @xmath154 .
the following algorithm uses the local approach of and to determine if an approximate solution corresponds to a real associated solution .
procedure @xmath155 : : input : : a real square polynomial system @xmath9 and an approximate solution @xmath3 with associated solution @xmath32 such that @xmath102 is invertible .
output : : a boolean @xmath156 that describes if @xmath148 .
begin : : do the following : + 1
. set @xmath157 . 2 .
if @xmath158 , * return * false . 3 .
if @xmath159 and @xmath160 , * return * true .
[ item : updatereal ] update @xmath161 , and return to ( a ) .
for the global approach to certifying real solutions , suppose that we know _ a priori _ that @xmath6 has exactly @xmath45 solutions .
suppose that @xmath162 are approximate solutions of @xmath6 with distinct associated solutions .
if , for all @xmath163 , @xmath164 and @xmath165 also correspond to distinct solutions , then @xmath166 and @xmath164 must correspond to the same solution , which is therefore real .
this global approach requires _ a priori _ knowledge about @xmath167 as well as approximate solutions corresponding to each solution to @xmath6 .
while it can not be applied to all systems , it is an alternative to the test based on @xmath37 . for a given set of points @xmath11 and a polynomial system @xmath6 , * certifysolns * , * certifydistinctsoln * , and * certifyrealsoln * answer the three questions posed in the introduction .
we provide a sketch of the algorithm .
procedure @xmath168 : : input : : a square polynomial system @xmath9 and a finite set of points @xmath169 such that if @xmath165 is an approximate solution with associated solution @xmath170 , then @xmath171 is invertible .
output : : a set @xmath172 consisting of certified approximate solutions to @xmath6 , a set @xmath173 consisting of points which have distinct associated solutions , and , if @xmath6 is a real map , a subset @xmath174 consisting of points which have real associated solutions .
begin : : -0.1 in + 1 .
. 2 . set @xmath176 and enumerate the points in @xmath177 as @xmath178 .
3 . for @xmath179 ,
set @xmath180 .
4 . for @xmath179 and for @xmath181 ,
if @xmath182 and @xmath183 are , set @xmath184 .
5 . set @xmath185 .
initialize @xmath186 .
if @xmath6 is a real polynomial system , do the following : 1 .
set @xmath187 and enumerate the points in @xmath188 as @xmath189 . 2 . for @xmath190 , if @xmath191 is , update @xmath192
when @xmath15 , the polynomial system @xmath14 is overdetermined . dedieu and
shub @xcite studied the overdetermined newton s method whose iterates are defined by @xmath193 where @xmath194 is the moore - penrose pseudoinverse of @xmath22 @xcite to determine conditions that guarantee quadratic convergence . since the fixed points of @xmath146
may not be solutions to the overdetermined polynomial system @xmath6 , this approach can not certify solutions to overdetermined polynomial systems .
we instead certify that points are associated solutions to two or more random square subsystems using the algorithms of section [ sec : square ] .
an additional level of security may be added by certifying that , for a given point which is an approximate solution to two or more random square subsystems , the associated solutions lie within a given distance of each other . as with the overdetermined newton s method ,
this also can not certify solutions to overdetermined polynomial systems , which is still an open problem .
let @xmath195 be a linear map , considered as a matrix in @xmath196 .
then @xmath197 gives a square polynomial system @xmath198 .
since @xmath199 for any @xmath200 , we call @xmath201 a _ square subsystem _ of @xmath6
. there is a nonempty zariski open subset @xmath202 such that for every @xmath203 and every @xmath39 , @xmath204 if and only if @xmath205 is invertible .
moreover , for every @xmath206 , @xmath205 is invertible .
see @xcite for more on square subsystems @xmath201 .
define @xmath207 which has dimension at most @xmath20 possibly passing through the origin .
a dimension - counting argument yields that there is a nonemtpy zariski open set @xmath208 such that , for every @xmath209 , @xmath210 is a linear space of dimension @xmath211 passing through the origin and @xmath212 .
in particular , if @xmath213 , then @xmath214 in addition , suppose that @xmath5 is an approximate solution to both @xmath215 and @xmath216 with associated solutions @xmath108 and @xmath109 , respectively . for @xmath26
, define @xmath217 for @xmath218 .
if @xmath111 , there exists @xmath26 such that @xmath219 certifying that @xmath220 .
if @xmath115 , then , for any @xmath18 , there exists @xmath26 such that @xmath221 certifying that @xmath222 .
in particular , this certifies that the solutions @xmath108 and @xmath109 to @xmath223 and @xmath224 associated to the common approximate solution @xmath5 lie within a distance @xmath19 of each other . for @xmath225 , this _ heuristically _ shows that @xmath115 . in summary ,
if @xmath5 is a certified approximate solution to two different square subsystems with distinct associated solutions , a certificate can be produced demonstrating this fact .
also ( but not conversely ) , for any given tolerance @xmath226 , a certificate can be produced that the distance between the associated solutions to the two square subsystems is smaller than @xmath19 .
an additional test using the function residual could be added to this process .
the following lemma describes such a test .
[ l : pie_in_sky ] let @xmath14 be an overdetermined polynomial system , @xmath227 , and @xmath5 be an approximate solution to @xmath228 with associated solution @xmath32 such that @xmath229 .
then there exists @xmath230 such that if there exists @xmath60 satisfying @xmath231 then @xmath232 .
define @xmath233 and @xmath234 .
we note that if @xmath235 , since @xmath236 , it is easy to verify that we can take @xmath237 hence , we can assume that @xmath238 .
since @xmath239 @xmath240
. moreover , theorem [ thm : alpharobust ] yields that @xmath241 .
assume @xmath242 .
since @xmath199 , @xmath243 .
in particular , @xmath244 is positive on the compact set @xmath245 .
thus , there exists @xmath230 such that @xmath246 for all @xmath247 . for lemma [ l : pie_in_sky ] to give an algorithm , we would need a general bound for the minimum of a positive polynomial on a disk . in cases when such a bound is known , e.g. , @xcite , it is too small to be practical .
the program * alphacertified * is written in c and depends upon gmp @xcite and mpfr @xcite libraries to perform exact rational and arbitrary precision floating point arithmetic .
when using rational arithmetic , all internal computations are _ certifiable_. because of the bit length growth of rational numbers under algebraic computations , * alphacertified * allows the user to select a precision and use floating point arithmetic in that precision to facilitate computations . since floating point errors from internal computations
are not fully controlled , * alphacertified * only yields a _ soft certificate _ when using the floating point arithmetic option .
when the polynomial system is overdetermined , * alphacertified * displays a message informing the user about what it has actually computed .
three input files are needed to run * alphacertified*. these files contain the polynomial system , the list of points to test , and the user - defined settings .
see @xcite for more details regarding exact syntax of these files .
the polynomial system is assumed to have rational complex coefficients and described in the input file with respect to the basis of monomials .
that is , the user inputs the coefficient and the exponent of each variable for each monomial term in each polynomial of the polynomial system .
the set of points to test are assumed to have either rational coordinates if using rational arithmetic or floating point coordinates if using floating point arithmetic .
when using floating point arithmetic , the points are inputted in the precision selected by the user .
the list of user - defined settings includes the choice between rational and floating point arithmetic , the floating point precision to use for the basic computations if using floating point arithmetic , and which certification algorithm to run .
the user can also define a value , say @xmath248 , such that , for each certified approximate solution , the associated solution will be approximated to within @xmath249 and printed to a file .
the specific output of * alphacertified * depends upon the user - defined settings . in each case , an on - screen table summarizes the output as well as a file that contains a human - readable summary for each point .
the other files created are machine - readable files that can be used in additional computations .
linear solving operations are performed using an @xmath250 decomposition and the spectral matrix norm is bounded above using the frobenius norm .
this choice further worsens the approximation of @xmath91 described in proposition [ prop : boundgamma ] , which has two direct consequences on the performance of the algorithms .
first , this requires that the value of @xmath251 must be smaller in order to certify approximate solutions in * certifysolns*. second , algorithms * certifydistinctsoln * and * certifyrealsoln * may need to utilize extra newton iterations .
apart from the added computational cost , the use of gmp and mpfr allows * alphacertified * to still perform these computations even when using such an approximation of @xmath91 . when using rational arithmetic , * alphacertified * avoids taking square roots when testing the required inequalities .
when using floating point arithmetic , as an effort to control the floating point errors , the internal working precision is increased when updating the point via a newton iteration , for instance in step ( d ) of * certifydistinctsoln * and step ( d ) of * certifyrealsoln*. the software * alphacertified * determines if a square polynomial system @xmath6 in @xmath20 variables is real using two tests .
the first test determines if the coefficients of @xmath6 are real .
the second selects a pseudo - random point @xmath252 and determines if @xmath253 .
the user either instructs * alphacertified * to bypass all tests and declare that @xmath6 is real , or which tests to use .
if all tests fail , then * alphacertified * bypasses the real certification .
otherwise , for each approximate solution @xmath5 with associated solution @xmath32 , * alphacertified * determines if there exists a real approximate solution that also corresponds to @xmath32 .
if the user incorrectly identified @xmath6 as real , then @xmath32 may not be real .
therefore , * alphacertified * displays a message informing the user about what it actually has certified . for an overdetermined polynomial system @xmath6 , *
alphacertified * only checks to see if all of the coefficients of @xmath6 are real . in this case ,
* alphacertified * randomizes @xmath6 using real matrices to obtain real square subsystems .
we used * alphacertified * to study four polynomial systems whose number of real solutions is relevant .
two are from kinematics and two are from enumerative geometry .
all involve polynomial systems that are not easily solved using certified methods from symbolic computation .
the files used in the computations , as well as instructions for their use , are found on our website @xcite .
computations of sections [ sec : lines ] and [ sec : schubert ] used nodes of the brazos cluster @xcite that consist of two 2.5 ghz intel xeon e5420 quad - core processors .
the stewart - gough platform is a parallel manipulator in which six variable - length actuators are attached between a fixed frame ( the ground ) and a moving frame ( the platform ) @xcite .
each position of the platform uniquely determines the lengths of the six actuators .
however , the lengths of the actuators do not uniquely determine the position and orientation of the platform , as there are typically several assembly modes , called _ positions_. a generic platform with generic actuator lengths has 40 complex assembly modes .
dietmaier @xcite used a continuation method to find a platform and leg lengths for which all 40 positions are real .
while his formulation as a system of polynomial equations and conclusions about their solutions being real have been reproduced numerically ( this is a problem in verschelde s test suite @xcite ) , these computations only give a heuristic verification of dietmaier s result .
we modified the polynomial system from verschelde s test suite , which uses the parameters obtained by dietmaier , by converting the floating point numbers to rational numbers .
we then ran phcpack @xcite on the resulting polynomial system to obtain 40 numerical solutions to the system , each of which it identified as real . after converting the floating point coordinates of the solutions to rational numbers
, we ran * alphacertified * using these rational polynomials and rational points .
it verified that these 40 points correspond to distinct real solutions .
this gives a rigorous mathematical proof of dietmaier s result .
a _ four - bar linkage _ is a planar linkage consisting of a triangle with two of its vertices connected to two bars , whose other endpoints are fixed in the plane .
the base of the triangle , the two attached bars , and the implied bar between the two fixed points are the four bars .
@xmath254 a general linkage has a one - dimensional constrained motion during which the joints may rotate , and the curve traced by the apex of the triangle is its _
workspace curve_. the _ nine - point path synthesis problem _ asks for the four - bar linkages whose workspace curve contains nine given points .
morgan , sommese , and wampler @xcite used homotopy continuation to solve a polynomial system which describes the four - bar linkages whose workspace curves pass through nine given points .
they found that for nine points @xmath255 in general position , there are 8652 isolated solutions .
due to a two - fold symmetry , there are 4326 distinct four - bar linkages which appear in 1442 triplets , called roberts cognates .
we used * alphacertified * to produce a soft certificate that the polynomial system has at least 8652 isolated solutions and , for a specific set of nine real points , certified the number of real solutions among these 8652 solutions . if @xmath256 , the formulation of @xcite is not a real polynomial system .
the usual approach of writing the variables using real and imaginary parts gives a real polynomial system @xmath257 consisting of four quadratic and eight quartic polynomials . for nine points @xmath255 ,
the polynomial system @xmath258 depends upon the variables @xmath259 define the complex numbers @xmath260 whose complex conjugates are @xmath261 , respectively .
these correspond to the variables used in the formulation in @xcite .
the four quadratic polynomials of @xmath257 are @xmath262 the eight quartic polynomials depend upon the displacements from @xmath263 to the other points @xmath264 . for @xmath265 , define @xmath266 and write each displacement @xmath267 using _ isotropic coordinates _ , namely @xmath268 where @xmath269 for @xmath265 , the quartic polynomial @xmath270 of @xmath257 is @xmath271 where @xmath272 and @xmath273 we first certified that , for nine randomly selected points in the complex plane , the resulting polynomial system has at least 8652 isolated solutions . since the displacements @xmath267
define the polynomial system , we choose them to be points of @xmath80 ^ 2 $ ] with each coordinate having unit modulus of the form @xmath274 where @xmath275 was a quotient of two ten digit random integers .
we used regeneration @xcite in bertini @xcite to compute 8652 points that were _ heuristically _ within @xmath276 of an isolated solution for @xmath257 .
then , * alphacertified * produced a soft certificate using 256-bit precision that these 8652 points are approximate solutions to @xmath257 with distinct associated solutions .
we next certified the number of real solutions for a specific set of nine real points , namely problem 3 of @xcite .
the nine real points are listed in table 2 of @xcite , which , for convenience , we list the values of @xmath277 in table [ tab : ninepts ] . 1 & 0.125 in @xmath278 ' '' '' + 2 & 0.125 in @xmath279 ' '' '' + 3 & 0.125 in @xmath280 ' '' ''
+ 4 & 0.125 in @xmath281 ' '' '' + 5 & 0.125 in @xmath282 ' '' ''
+ 6 & 0.125 in @xmath283 ' '' '' + 7 & @xmath284 ' '' '' + 8 & @xmath285 ' '' '' + since the points are real , @xmath286 is the conjugate of @xmath277 .
we used parameter continuation in bertini to solve the resulting polynomial system starting from the 8652 solutions to the polynomial system solved in the first test .
this generated a list of 8652 points which * alphacertified * soft certified using 256-bit precision to be approximate solutions that have distinct associated solutions of which 384 are real .
in particular , this confirms the results reported in table 3 of @xcite for problem 3 , namely , that @xmath287 of the 1442 mechanisms are real .
figure [ f : four - bar ] shows three of the 64 real mechanisms that solve this synthesis problem , together with their workspace curves .
the first has two assembly modes with the workspace curve of one mode a simple closed curve that contains the nine target points .
this mechanism is the only viable mechanism among the 64 real mechanisms .
the second has only one assembly mode , but its workspace curve is convoluted and does not meet the target points in a useful order .
the third has two assembly modes , and each only reaches a proper subset of the target points .
we consider geometric problems of plane conics in @xmath288 that meet @xmath45 points and @xmath289 lines for @xmath290 . when the points and lines are general , the numbers of plane conics are known and presented in table [ t : plc ] .
.numbers of plane conics [ cols=">,>,>,>,>,>",options="header " , ] this computation was part of a larger test of hypothesized lower bounds @xcite .
smale s @xmath0-theory provides a way to certify solutions to polynomial systems , determine if two points correspond to distinct solutions , and determine if the corresponding solution is real . using either exact rational or arbitrary precision floating point arithmetic ,
* alphacertified * is a program which implements these @xmath0-theoretical methods .
we have also produced a maple interface to * alphacertified * to facilitate the construction of the input files needed .
the authors would like to thank mike shub for his helpful comments and the first author would like to thank the organizers of the foundations of computational mathematics thematic program at the fields institute .
bates , j.d .
hauenstein , c. peterson , and a.j .
sommese , _ numerical decomposition of the rank - deficiency set of a matrix of multivariate polynomials _ , approximate commutative algebra , texts monogr .
comput . , springer , vienna , 2009 , pp .
5577 .
l. fousse , g. hanrot , v. lefvre , p. plissier , and p. zimmermann , _ mpfr : a multiple - precision binary floating - point library with correct rounding _ , acm trans .
software * 33 * ( 2007 ) , no . 2 , art . 13 , 15 . s. smale , _
newton s method estimates from data at one point _ , the merging of disciplines : new directions in pure , applied , and computational mathematics ( laramie , wyo . , 1985 ) , springer , new york , 1986 , pp . 185196
sommese , j. verschelde , and c.w .
wampler , _ numerical irreducible decomposition using projections from points on the components _ , symbolic computation : solving equations in algebra , geometry , and engineering ( south hadley , ma , 2000 ) , contemp .
286 , amer .
soc . , providence , ri , 2001 , pp .
3751 .
j. verschelde , _ algorithm 795 : phcpack : a general - purpose solver for polynomial systems by homotopy continuation _ , acm trans .
* 25 * ( 1999 ) , no . 2 , 251276 , software available at http://www.math.uic.edu/~jan . | smale s @xmath0-theory uses estimates related to the convergence of newton s method to certify that newton iterations will converge quadratically to solutions to a square polynomial system . the program
* alphacertified * implements algorithms based on @xmath0-theory to certify solutions of polynomial systems using both exact rational arithmetic and arbitrary precision floating point arithmetic .
it also implements algorithms that certify whether a given point corresponds to a real solution , and algorithms to heuristically validate solutions to overdetermined systems .
examples are presented to demonstrate the algorithms . |
let @xmath3 be a knot or link in @xmath4 .
the knot quandle @xmath5 is an algebraic knot invariant that can be derived from a diagram of @xmath3 . in @xcite , @xcite
, joyce proves that the knot quandle is a complete invariant of unoriented knots .
however , as with knot groups , it is easy to obtain a presentation of the quandle , but hard to determine if two presentations yield isomorphic quandles . passing to the @xmath6-quandle , @xmath7 , a certain quotient of @xmath5 , yields a more tractable knot invariant about which much is still unknown .
this paper is one piece of a larger project to tabulate all links with finite @xmath6-quandles and furthermore to explicitly describe those quandles .
we hope to complete this project in future publications . combining results of joyce @xcite , @xcite , and winker @xcite
, it follows that if @xmath7 is finite , then @xmath8 is finite , where @xmath9 is the the @xmath6-fold cyclic branched cover of @xmath10 , branched over @xmath3 . in this paper
, we examine the class of montesinos links @xmath0 , which we call @xmath1-montesinos links , and which are known to have 2-fold cyclic branched covers with finite fundamental group .
we show that their 2-quandles , also known as _
quandles , are in fact finite .
moreover , we carefully describe the nature of these quandles .
our main result is the following theorem .
[ main theorem ] if @xmath0 is a @xmath1-montesinos link with @xmath11 , + @xmath12 , and @xmath13 , then the order of @xmath14 is @xmath15 . in section [ quandles ] , we review basic material about involutory quandles and describe the diagramming method introduced by winker @xcite used to find a cayley graph of a quandle .
then , in section [ 22rmontesinos ] we describe the @xmath1-montesinos links and derive a presentation for their involutory quandles .
we then employ winker s method in section [ cayley graphs ] to compute the involutory quandles of these links , describing the quandle by means of its cayley graph . finally , in section [ properties ] , we describe several properties of these quandles .
in this section we quickly review the definition of the fundamental quandle of a link and its associated involutory quandle .
we refer the reader to @xcite , @xcite , @xcite , and @xcite for more detailed information .
a _ quandle _ is a set @xmath16 together with two binary operations @xmath17 and @xmath18 which satisfies the following three axioms .
* @xmath19 for all @xmath20 . * @xmath21 for all @xmath22 . *
@xmath23 for all @xmath24 .
if @xmath25 is an oriented knot or link in @xmath4 , then a presentation of its _ fundamental quandle _ , @xmath26 , can be derived from a regular diagram @xmath27 of @xmath25 .
this process mimics the wirtinger algorithm .
namely , assign a quandle generator @xmath28 to each arc of @xmath27 , then at each crossing introduce a relation as follows .
suppose the overcrossing strand is labeled @xmath29 while the undercrossing strand to the left ( using the orientation of the overcrossing strand ) is labeled @xmath30 and the undercrossing strand to the right is labeled @xmath31 .
then @xmath32 .
it is easy to check that the three riedemeister moves do not change the quandle given by this presentation so that the quandle is indeed an invariant of the oriented link .
it is important to note that the operation @xmath17 is , in general , not associative . throughout this paper
, we will assume that any expression without parentheses is to be grouped in left - associated form .
for example , the expression @xmath33 should be taken to be @xmath34 .
it is an interesting exercise to show that the quandle axioms can be used to rewrite every product in left - associated form .
thus , there is no loss of generality in working entirely with left - associated products .
a quandle is called _ involutory _ if @xmath35 . in an involutory quandle axiom a2
becomes + @xmath36 for all @xmath22 .
the involutory quandle for a knot or link is presented in the same manner as its fundamental quandle except that the orientation of the link is now irrelevant .
the involutory quandle of a link @xmath25 will be denoted by @xmath14 .
the involutory quandle of a link may be thought of as a quotient of the fundamental quandle and in general one should expect a significant loss of information in passing from @xmath26 to @xmath14 .
all quandles considered in this paper will be involutory quandles .
finally , we adopt the exponential notation used in @xcite , denoting @xmath37 as @xmath38 .
this has the advantage of further clarifying the ambiguity caused by lack of associativity .
now @xmath39 will be taken to mean @xmath40 whereas @xmath41 will mean @xmath42 .
the following lemma from @xcite , which describes how to re - associate a product in an involutory quandle , will be used repeatedly in this paper . if @xmath43 are elements of an involutory quandle , then @xmath44 [ leftassoc ] one of the main tools we use in this paper is winker s method to produce the cayley graph of an involutory quandle from a given presentation @xcite .
the method is similar to the well - known todd - coxeter method of enumerating cosets of a subgroup of a group @xcite .
suppose @xmath14 is presented as @xmath45 where each @xmath46 is a word in the generators .
if @xmath47 is any element of the quandle , then it follows from lemma [ leftassoc ] that @xmath48 where @xmath49 is the word @xmath50 written backwards .
winker calls this relation the _ secondary relation _ associated to the _ primary relation _ @xmath51 .
winker s method now proceeds as follows . 1 .
begin with one vertex for each generator @xmath30 , labeled with the integer @xmath52 .
2 . add a loop at each vertex @xmath30 , labeled @xmath30 .
( this encodes axiom a1 . )
3 . using each primary relation @xmath51 ,
in their given order , introduce vertices and edges as necessary to create the path from @xmath53 to @xmath54 given by @xmath46 .
this process is called _ tracing _ the primary relation .
vertices are labeled with consecutive integers in the order they are introduced and edges are labeled with the corresponding generator .
tracing the @xmath52-th relation may introduce like - labeled edges sharing a vertex which we now identify , possibly leading to other necessary identifications .
this process is called _ collapsing _ and all collapsing is carried out before tracing the @xmath55-st relation .
4 . proceeding in order through the list of vertices , trace and collapse each secondary relation ( in order ) .
all secondary relations are traced and collapsed at vertex @xmath52 before proceeding to vertex @xmath56 .
if the method terminates in a finite graph , then this is the cayley graph of the quandle .
however , the method may not terminate .
the reader is referred to winker @xcite for more details .
for example , let @xmath3 be the the standard 3-crossing diagram of the right - handed trefoil knot . if the arcs are labeled @xmath57 , and @xmath58 in order as we traverse the knot , then we obtain the presentation @xmath59 of its involutory quandle .
as explained in @xcite , presentations can be changed by the analog of tietze transformations .
here we can eliminate @xmath58 obtaining @xmath60 to which we now apply winker s method .
after the first two steps , we arrive at the graph shown in figure [ trefoil ] . tracing the single secondary relation , @xmath61 at each of the vertices ( in order ) introduces no additional vertices and requires no collapsing .
therefore , this is the cayley graph of the involutory quandle of the right - handed trefoil knot .
consider the unoriented montesinos link @xmath62 , shown in figure [ generalmontesinos ] , where @xmath63 , @xmath64 , and @xmath65 . here
the box labelled @xmath66 denotes the rational tangle determined by the fraction @xmath66 and @xmath67 is the number of right - handed half - twists . by making use of flypes
, we may change each numerator @xmath68 by multiples of @xmath69 at the expense of changing @xmath67 .
in particular , if some @xmath68 is replaced with @xmath70 , then we replace @xmath67 with @xmath71 .
hence by allowing arbitrary values for @xmath67 , we may assume that @xmath72 for each tangle .
in addition , cyclic permutation of the tangles together with rotation around a vertical axis allows us to assume @xmath73 .
the reader is directed to @xcite for more information on montesinos links . for the remainder of the paper we consider the link @xmath74 .
we begin by deriving a presentation for the involutory quandle @xmath14 . ]
we illustrate this process with the link @xmath75 shown in figure [ montesinoslink ] .
label the arcs of the diagram with quandle elements 1 , 2 , and 3 as shown in the figure .
( these might more properly be called @xmath57 , and @xmath58 , but we simply write 1 , 2 , 3 . ) the elements 1 , 2 , and 3 generate the quandle . notice that
for all @xmath76 , we obtain the relation @xmath77 . however , there are two more relations that depend on both @xmath67 and @xmath76
. first note that by moving from left to right through the @xmath67 right - handed half - twists in the upper part of the diagram , the two arcs to the right receive labels of @xmath78 and @xmath79 where @xmath80 and @xmath81 are some words in the generators 1 and 2 that depend on @xmath67 .
next , if we follow the strand labeled @xmath82 into the @xmath76 tangle , we will come out at one of the other four endpoints of the tangle and obtain a second relation .
similarly , following the other string in this 2-string tangle gives the third relation . ]
we begin by determining @xmath80 and @xmath81 , which are easily found by induction on @xmath67 . in the following lemma , and throughout the paper ,
if @xmath83 is a word in the generators , then we use the notation @xmath84 to denote @xmath83 written backwards .
additionally , we use @xmath85^n$ ] to stand for @xmath83 repeated @xmath6 times when @xmath86 , to stand for the empty string when @xmath87 , and to stand for @xmath88^{-n}$ ] when @xmath89 . the words @xmath80 and @xmath81 depend on @xmath67 and
are given as , @xmath90^{e-2}2 $ ] and @xmath91^{e-1}2 $ ] .
[ ab ] we now consider the @xmath76 tangle .
there are three different cases : @xmath92 even and @xmath93 odd , @xmath92 odd and @xmath93 even , and finally , @xmath92 odd and @xmath93 odd .
the montesinos link has a total of three components if @xmath92 is even and two if @xmath92 is odd .
moreover , the parity of @xmath93 and @xmath92 determine which pairs of endpoints are connected by the two strings of the tangle .
if @xmath92 is even , then the lower left endpoint is connected to the upper left endpoint .
if @xmath92 is odd , and @xmath93 is even , then the lower left endpoint is connected to the lower right endpoint . finally ,
if both are odd , the lower left endpoint is connected to the upper right endpoint . and @xmath92 both odd , shown with @xmath94 . ]
we describe the case with @xmath93 and @xmath92 both odd .
after isotopy , the tangle can be represented as shown in figure [ pqtangle ] .
consider the lower left endpoint of the tangle .
this arc is labeled @xmath82 .
as we move along this arc , we alternately pass under arcs labelled @xmath78 or @xmath79 and @xmath82 .
we pass under the arc labelled @xmath82 a total of @xmath95 times , under @xmath78 a total of @xmath96 times , and under @xmath79 a total of @xmath97 times .
this path ends at @xmath78 yielding a relation of the form @xmath98 . by lemmas [ leftassoc ] and [ ab ]
, @xmath83 is an alternating product of the words @xmath99 or @xmath100 and @xmath101 .
for example , with @xmath94 as shown in figure [ pqtangle ] we obtain the relation @xmath102 for any @xmath76 , we can simplify the relation using the following result , the proof of which we postpone to the next section .
[ simplify relation ] for all @xmath103 , @xmath104 .
repeated application of this lemma allows us to collect all @xmath105 and @xmath106 expressions together and so the relation @xmath107 can be rewritten as @xmath108^{(p-1)/2}[b1b232]^{(q - p)/2}}=1^a.\ ] ] now applying this same process starting at the arc labelled @xmath109 gives a second relation @xmath110^{(p+1)/2}[b1b232]^{(q - p-2)/2}}=1^b.\ ] ] finally , a similar analysis for the other parities of @xmath93 and @xmath92 gives the following presentation for the involutory quandle .
the reader is warned that , in an effort to make the relations appear more uniform across the cases , we introduce some canceling generators to rewrite the last case . [ r2r3 ] if @xmath25 is the montesinos link @xmath111 with @xmath11 , then the involutory quandle of @xmath25 can be presented as @xmath112 , where @xmath113 is the relation @xmath77 . the relations @xmath114 and @xmath115 depend on the parity of @xmath93 and @xmath92 as follows : -.35 in @xmath116^{(q - p)/2}[232a1a]^{(p-1)/2}2 } & 3=1^{b [ 232 b1 b]^{(q - p-2)/2}[232a1a]^{(p+1)/2}}\\ \text{even}&\text{odd } & 3=1^{b[232b1b]^{(q - p-1)/2}[232a1a]^{p/2}2 } & 3=1^{a[232b1b]^{(q - p-1)/2}[232a1a]^{p/2}}\\ \text{odd}&\text{even } & 3=3^{2 [ 232b1b]^{(q - p-1)/2 } [ 232a1a]^{(p+1)/2 } } & 1=1^{a[232b1b]^{(q - p+1)/2}[232 a1a]^{(p-1)/2}b}\\ \end{array}\ ] ]
in this section we explicitly construct the cayley graph of @xmath117 with respect to the generators 1 , 2 , and 3 , given in lemma [ r2r3 ] .
let @xmath118 be the component of the cayley graph containing the generator @xmath52 .
each component @xmath118 is best described as the quotient of an infinite , labeled , planar graph @xmath119 which we now describe .
start with a single rectangular tile one unit wide and four units tall with its lower left corner located at the point @xmath120 .
consider this rectangle as a graph with six edges and six vertices located at @xmath120 , @xmath121 , @xmath122 , @xmath123 , @xmath124 , @xmath125 .
let @xmath126 be the union of all translates of this tile obtained by shifting @xmath127 units horizontally and @xmath6 units vertically , where @xmath127 and @xmath6 are integers with @xmath128 .
label all horizontal edges with 2 , and all vertical edges whose upper endpoint is located at @xmath129 with 3 if @xmath130 ( mod 4 ) and with 1 if @xmath131 ( mod 4 ) .
finally , let @xmath132 be obtained from @xmath133 by labeling the vertex @xmath120 with generator @xmath52 .
the graph @xmath134 admits a number of label - preserving symmetries that we will use to describe @xmath14 . for any integers @xmath6 and @xmath127 ,
let @xmath135 denote the @xmath136-radian rotation centered at @xmath137 , let @xmath138 denote the translation @xmath139 , let @xmath140 denote reflection across the line @xmath141 , and let @xmath142 denote the glide reflection @xmath143 .
it is easy to check that each of these transformations is a label - preserving automorphism of @xmath133 . if @xmath144 are label - preserving automorphisms of @xmath133 we denote by @xmath145 the group of automorphism which they generate and by @xmath146 the quotient graph .
[ cayley graph ] suppose @xmath147 and let @xmath148 .
then the components of the cayley graph of @xmath14 are : 1 . if @xmath92 is odd , then @xmath149 and @xmath150 .
2 . if @xmath92 is even , then @xmath151 , @xmath150 , and @xmath152 .
the component @xmath153 is obtained by trading all labels of 1 and 3 in @xmath154 .
* remark : * it follows from theorem [ cayley graph ] that a fundamental domain @xmath155 for @xmath154 can be taken to be the subset of @xmath156 contained in @xmath157\times[0,4q]$ ] if @xmath92 is odd and contained in @xmath158 \times [ 0 , 2q]$ ] if @xmath92 is even . if @xmath92 is odd , @xmath155 is not quite a fundamental domain , as pairs of vertices symmetric with respect to @xmath159 or with respect to @xmath160 still need to be identified .
a fundamental domain @xmath161 for @xmath162 , in either case , is the subset of @xmath163 contained in @xmath164\times[0,4]$ ] . counting the number of vertices in @xmath155 and @xmath161
gives the following corollary .
[ corollary]the orders of the components of the cayley graph of @xmath14 are : 1 .
if @xmath92 is odd , then @xmath165 and @xmath166 .
2 . if @xmath92 is even , then @xmath167 and @xmath166 . in both cases , @xmath168 .
notice that theorem [ main theorem ] now follows immediately from theorem [ cayley graph ] and its corollary . for example
, in the case where @xmath169 and @xmath170 , @xmath155 , @xmath154 , and @xmath162 are shown in figures [ fund domain ] , [ c1 q odd ] , and [ c2 ] . in figure [ c1 q even ] , we illustrate the case of @xmath171 with @xmath172 . of @xmath14 for @xmath173 , @xmath174 , and @xmath170 .
] of @xmath14 for @xmath173 , @xmath174 , and @xmath170 . here
blue , black , and red correspond to the generators 1 , 2 , and 3 , respectively . ] of @xmath14 for @xmath173 , @xmath174 , and @xmath170 . here
blue , black , and red correspond to the generators 1 , 2 , and 3 , respectively . ] of @xmath14 for @xmath173 , @xmath175 , and @xmath172 . here
blue , black , and red correspond to the generators 1 , 2 , and 3 , respectively .
letters indicate how half - edges at the right and left join . ]
our overall strategy for proving these theorems is to apply the diagramming method of winker @xcite to a presentation of the quandle @xmath14 . in order to simplify the application of winker s method
, we will first pass to an equivalent presentation of the quandle .
our approach is motivated by first considering the involutory quandle @xmath176 .
imposing additional relations on @xmath177 cause vertices in its cayley graph to be identified , thus leading to the cayley graph of @xmath14 . applying winker s method to @xmath177 ,
the graph @xmath178 arises in a natural way because the primary relation @xmath77 gives rise to the secondary relation @xmath179 for every element @xmath180 in @xmath177 .
thus it is not surprising to see the hexagons of @xmath178 appearing in the end result .
in fact , winker s method results in three components : @xmath181 starting with the graphs @xmath156 and @xmath163 , and applying the relations @xmath114 and @xmath115 in lemma [ r2r3 ] , we were led to theorem [ cayley graph ] .
however , we will now treat this approach as simply providing guidance to our formal proof of the theorem .
using @xmath178 , we will see how to rewrite the presentation of @xmath14 in such a way that applying winker s method to the new presentation will produce the cayley graphs described in theorem [ cayley graph ] after tracing and collapsing the primary relations .
each of the secondary relations will then hold without further modification of the graph .
we begin with some general observations about tracing a word in the generators starting at various vertices in @xmath178 .
first notice that there are four _ types _ of vertices @xmath129 in @xmath134 determined by the residues @xmath180 ( mod 2 ) and @xmath47 ( mod 4 ) .
we denote these four vertex types in the obvious way as * 01 * , * 03 * , * 11 * , and * 13*. given a word @xmath83 in the generators 1 , 2 , and 3 , we can start at any vertex @xmath182 in @xmath178 , trace the word @xmath83 , and arrive at a vertex @xmath183 .
it is easy to see that the vertex type of @xmath183 depends only on @xmath83 and the type of @xmath182 .
thus , each word @xmath83 determines a permutation of the set of four types of vertices .
let @xmath184 denote this permutation .
for example , @xmath185 is the product of two transpositions , written in cycle notation as @xmath186 .
if tracing the word @xmath83 starting at vertex @xmath182 in @xmath178 ends in vertex @xmath183 , then let @xmath187 denote the vector @xmath188 in the plane .
we call @xmath187 the _ displacement _ of @xmath83 starting at @xmath182 . because of the translational symmetry taking any vertex to another one of the same type
, it follows that @xmath187 depends only on @xmath83 and the type of @xmath182 .
therefore , we will often write one of * 01 * , * 03 * , * 11 * , or * 13 * in place of @xmath182
. the following lemma will be helpful in determining displacements .
the proof is left to the reader .
[ displacements ] let @xmath83 be any word in the generators @xmath189 , @xmath190 , and @xmath109 .
if @xmath191 , then 1 .
@xmath192 2 .
@xmath193 equals @xmath194 , @xmath195 , @xmath196 , or @xmath197 , if @xmath184 equals @xmath198 , @xmath199 , @xmath185 , or @xmath200 , respectively .
lemma [ displacements ] implies that if the words @xmath83 and @xmath201 have the same displacement starting at some vertex @xmath182 , then they have the same displacement starting at any vertex .
if this is the case , we say that @xmath83 and @xmath201 _ have the same displacement_. using lemma [ displacements ] , we may compute the permutations and displacements of the words that appear in the relations @xmath114 and @xmath115 . again
, details are left to the reader .
[ specific displacements ] the permutation and displacement of @xmath80 , @xmath81 , @xmath202 , and @xmath203 are : @xmath204 let us now focus on the case where both @xmath93 and @xmath92 are odd .
the component @xmath205 is the quotient @xmath206 . the pair of vertices @xmath120 and @xmath207 both represent the generator 1 .
the relation @xmath114 is of the form @xmath208 where @xmath209^{(q - p)/2}[232a1a]^{(p-1)/2}2 $ ] .
thus , if we start at the vertex @xmath120 and trace the word @xmath210 , we will arrive at a vertex @xmath183 that is equal to the generator 3 . to determine the coordinates of the vertex @xmath183 , we use lemmas [ displacements ] and [ specific displacements ] to calculate the displacement of @xmath210 starting at @xmath120 .
assume that @xmath67 is even .
then , by lemma [ specific displacements ] , we have @xmath211 and @xmath212 .
so , starting at @xmath120 and following @xmath80 , we end at vertex @xmath213 which has type @xmath214 .
this observation together with the fact that @xmath215 and @xmath216 are the identity then gives @xmath217 thus , the vertex @xmath218 is equal to the generator 3 .
since @xmath93 and @xmath92 are odd and @xmath67 is even this point is of type * 01*. moving from this vertex along the edge labelled @xmath109 we arrive at the type @xmath219 vertex @xmath220 .
however , because @xmath221 , it follows that the vertex @xmath220 also represents the generator 3 .
if @xmath67 is odd , then a similar calculation shows that @xmath222 and that the vertices @xmath223 represent the generator 3 .
the relation @xmath115 is of the form @xmath224 where @xmath225^{(q - p-2)/2}[232a1a]^{(p+1)/2}$ ] . calculating the displacement of @xmath226 yields @xmath227 if @xmath67 is even and @xmath228 if @xmath67 is odd .
therefore , regardless of the parity of @xmath67 , we have that the vertices @xmath229 and @xmath223 all represent the generator 3 .
it now follows that any two vertices that are symmetric with respect to either @xmath230 or @xmath231 represent equivalent elements in @xmath14 .
thus these points become rotational centers of symmetry .
similar calculations give the displacements of the words @xmath210 and @xmath226 that appear in relations @xmath114 and @xmath115 for the other parity cases .
we summarize the results in table [ r2r3 displacements ] .
lllll @xmath93&@xmath92&@xmath114&@xmath115&@xmath232 + + odd&odd&@xmath208&@xmath224&@xmath233 + & & & & @xmath234 + + even&odd&@xmath208&@xmath224&@xmath235 + & & & & @xmath236 + + odd&even&@xmath237&@xmath238&@xmath239 + & & & & @xmath240 + [ r2r3 displacements ] as already mentioned , we plan to change the presentation of @xmath14 .
finding an equivalent presentation for @xmath14 will depend heavily on the following lemma .
[ path homotopy ] if @xmath83 and @xmath201 are words in the generators @xmath241 having the same displacement , then the relation @xmath113 implies that @xmath242 for all @xmath243 .
* proof : * because @xmath77 , it follows that @xmath244 or equivalently that @xmath179 for any element @xmath180 in @xmath177 . more generally , suppose @xmath245 and @xmath246 are words in the generators 1 , 2 , and 3 such that their concatenation @xmath247 is a cyclic permutation of 121323 or its reverse .
then , for all @xmath180 and all words @xmath210 and @xmath226 , the relations @xmath248 and @xmath249 are equivalent .
this last statement can be interpreted in terms of the graph @xmath119 by seeing that we are moving the path @xmath250 across one of the hexagonal faces to reach the path @xmath251 .
call such a move an _
hexagonal move_. because @xmath83 and @xmath201 have the same displacement , the paths starting at @xmath180 that they each determine clearly end at the same vertex @xmath47 .
moreover , we may move one path to the other by a sequence of hexagonal moves .
@xmath252 it is not difficult to show that @xmath253 and @xmath254 have the same displacement .
thus , lemma [ simplify relation ] , the proof of which we had postponed , now follows easily from lemma [ path homotopy ] .
we are now in a position to derive a new presentation of @xmath14 using lemma [ path homotopy ] .
the new presentation will be defined in terms of the following sets of relations .
recall that @xmath148 and notice that @xmath113 is equivalent to @xmath255 .
@xmath256^{|w|}[13]^{(q-2)/2 } } & \gamma ' : 3&=3^{[12]^{|w|}[31]^{(q-2)/2}}\\ \delta_i : 1&=1^{[21]^i [ 31]^q[12]^i } & \delta ' : 3&=3^{[13]^q } \\ \epsilon_i : 1&=1^{[23]^i[13]^{\pm q/2}[23]^{|w|-i}3 } & \epsilon_i ' : 3&=3^{[12]^i[13]^{\pm q/2}[12]^{|w|-i}1}\\ \zeta_{ik } : 1&=1^{[23]^i[31]^k21[31]^{k-1}[32]^{i+1 } } & \zeta_{ik } ' : 3&=3^{[12]^i[31]^k32[31]^{k+1}[21]^{i+1}}\\ \eta_k : 1&=1^{[23]^{(|w|-1)/2}[31]^k2[13]^{k+q/2}[32]^{(|w|-1)/2 } } & \eta'_k : 3&=3^{[12]^{(|w|-1)/2}[13]^k121[31]^{k+q/2}[21]^{(|w|-1)/2}}\\ \theta_{ik } : 1&=1^{[21]^i[31]^k21[31]^k[12]^{i+1 } } & \mu : 3&=1^{[21]^{|w|}[31]^{(q-1)/2}}\\ \iota_i : 2&=2^{[12]^i31[21]^i } & \kappa : 2&=2^{[12]^{|2w|}}\\\end{aligned}\ ] ] [ new presentation ] let @xmath148 . then @xmath257 where @xmath258 , @xmath259 , @xmath260 , @xmath261 , and @xmath262 .
* proof : * there are several cases to consider .
assume first that @xmath92 is odd and @xmath263 .
we first show that that @xmath264 and @xmath265 can be derived from @xmath266 , and @xmath115 and vice versa . from table [ r2r3 displacements ] , we have the relations @xmath114 and @xmath115 are @xmath208 and @xmath224 , respectively .
regardless of the parity of @xmath93 and @xmath67 , using lemma [ displacements ] one can show that the displacement of @xmath267^q$ ] is the same as that of @xmath268 and that the displacement of @xmath269^{|w|}[31]^{(q-1)/2}$ ] is the same as that of @xmath210 .
therefore , by lemma [ path homotopy ] we have @xmath270^q}=3^{\overline{\beta}\alpha 3}$ ] and @xmath271^{|w|}[31]^{(q-1)/2}}=1^\alpha$ ] .
note that these two relations only depend on relation @xmath113 .
therefore , we have : @xmath272^{|w| } [ 31]^{(q-1)/2 } } & & \mbox{$\mu$ is a consequence of $ r_1 $ and $ r_2 $ } \\
3=3 ^ 3 \stackrel{r_2}{=}1^{\alpha 3 } \stackrel{r_3}{= } 3^{\overline{\beta}\alpha 3 } \stackrel{r_1}{=}3^{[13]^q } & & \mbox{$\delta'$ is a consequence of $ r_1 , r_2 $ , and $ r_3 $ } \\
1^\alpha \stackrel{r_1}{= } 1^{[21]^{|w| } [ 31]^{(q-1)/2 } } \stackrel{\mu}{=}3 & & \mbox{$r_2 $ is a consequence of $ r_1 $ and $ \mu$ } \\
1^\beta \stackrel{r_2}{= } 3^{\overline{\alpha}\beta}=3^{3\overline{\alpha}\beta } \stackrel{\delta'}{= } 3^{[13]^q3\overline{\alpha}\beta } \stackrel{r_1}{=}3 & & \mbox{$r_3 $ is a consequence of $ r_1 $ , $ \mu$ , and $ \delta'$ } \end{array}\ ] ] assume next that @xmath92 is odd and @xmath273 . in this case , the displacement of @xmath267^q$ ] is still the same as that of @xmath268 but the displacement of @xmath269^{|w|}[31]^{(q-1)/2}$ ] is now the same as that of @xmath274 .
so , in this case , we use the consequences @xmath270^q}=3^{\overline{\beta}\alpha 3}$ ] and @xmath271^{|w|}[31]^{(q-1)/2}}=1^\beta$ ] of @xmath113 .
@xmath275^{|w| } [ 31]^{(q-1)/2 } } & & \mbox{$\mu$ is a consequence of $ r_1 $ and $ r_3 $ } \\
3=3 ^ 3 \stackrel{r_2}{=}1^{\alpha 3 } \stackrel{r_3}{= } 3^{\overline{\beta}\alpha 3 } \stackrel{r_1}{=}3^{[13]^q } & & \mbox{$\delta'$ is a consequence of $ r_1 , r_2 $ , and $ r_3 $ } \\
1^\beta \stackrel{r_1}{= } 1^{[21]^{|w| } [ 31]^{(q-1)/2 } } \stackrel{\mu}{=}3 & & \mbox{$r_3 $ is a consequence of $ r_1 $ and $ \mu$ } \\ 1^\alpha \stackrel{r_3}{= } 3^{\overline{\beta}\alpha } \stackrel{r_1}{= } 3^{[13]^q 3 } \stackrel{\delta'}{= } 3 ^ 3 = 3 & & \mbox{$r_2 $ is a consequence of $ r_1 $ , $ \mu$ , and $ \delta'$ } \end{array}\ ] ] next we will show that @xmath276 , and @xmath277 can all be derived from @xmath278 , and @xmath279 . because @xmath114 and @xmath115 can be derived from @xmath278 , and @xmath279 , we are free to use these relations as well . to derive the relations @xmath280 , first note that @xmath269^i[31]^q[12]^i$ ] and @xmath281 have the same displacement if @xmath52 is even .
thus @xmath271^i[31]^q[12]^i}=1^{\alpha 3 \overline{\beta } } = 3^{3 \overline{\beta}}=3^{\overline{\beta}}=1 $ ] .
if @xmath52 is odd , then @xmath269^i[31]^q[12]^i$ ] and @xmath282 have the same displacement and a similar argument shows that @xmath283 is a consequence of @xmath266 , and @xmath115 .
it is not hard to show that for all @xmath284 and all @xmath285 , the displacement of @xmath269^i [ 31]^j 21[31]^j[12]^{i+1}$ ] is @xmath286 .
thus the relations @xmath287 all follow from @xmath113 .
the word @xmath288^i31[21]^i$ ] has the same displacement as @xmath289 if @xmath52 is even and as @xmath290 if @xmath52 is odd .
thus , when @xmath52 is even , we have @xmath291^i31[21]^i}$ ] , and when @xmath52 is odd , @xmath292^i31[21]^i}$ ] . therefore the relations @xmath293 are all consequences of @xmath113 .
a similar argument actually shows that if @xmath294 is any vertex in @xmath163 , then @xmath295 in @xmath296 . to see this , note that for all @xmath83
, @xmath297 has the same displacement as either @xmath290 or @xmath289 .
so , either @xmath298 thus any two vertices in @xmath163 separated by a vertical translation of @xmath299 units for any integer @xmath300 represent equal elements in @xmath296 .
finally , we can easily compute that @xmath301^{|2w|}=\langle -|2w| , 0 \rangle$ ] .
moreover , the primary relation @xmath208 implies that @xmath302 . using lemma [ displacements ]
, it follows that @xmath303 .
hence , letting @xmath83 be either @xmath304 or its reverse @xmath305 , we may assume @xmath306 and hence @xmath307^{\pm q}=\langle -|2w| , 0 \rangle$ ] . therefore , @xmath308^{|2w|}}=2^{x[13]^{\pm
q}}=2^{[13]^{\pm q}}=2 $ ] because @xmath267^{q}$ ] represents vertical translation by a multiple of 4 .
thus @xmath309 is a consequence of @xmath113 and @xmath114 .
this proves the lemma in the case where @xmath92 is odd .
we leave the proof in the case where @xmath92 is even to the reader .
@xmath252 * proof of theorem [ cayley graph ] : * we will now apply winker s method to the presentation in lemma [ new presentation ] , focussing first on the case where @xmath92 is odd . as we follow the method ,
imagine filling in vertices and edges in the graphs @xmath154 and @xmath162 already described in figures [ c1 q odd ] and [ c2 ] .
we start with the introduction of three vertices representing the generators 1 , 2 , and 3 , and each with a loop attached due to the idempotentcy relations .
next , the primary relations are used to add additional vertices and edges , if necessary .
moreover , we apply these in the following order : @xmath310 , and @xmath293 .
tracing @xmath265 connects the generator 1 to the generator 3 along a horizontal section ( a path of edges @xmath269^{|w|}$ ] ) followed by the right vertical side .
next , tracing @xmath311 closes the loop at the lower right corner .
the relation @xmath312 fills in the left vertical side .
the relations @xmath283 , for @xmath313 , now leave the horizontal section , fill in the vertical sections , and then return to the horizontal section . finally , the relations @xmath314 fill in all the remaining horizontal edges labeled 2 .
the component @xmath315 is now complete .
the relation @xmath309 traces out the horizontal section of @xmath162 and finally , the relations @xmath293 fill in the vertical sections , thus completing @xmath162 .
the next step in winker s method is to apply the secondary relations to each already existing vertex . in general , when checking a secondary relation for a given vertex @xmath180 , it may be necessary to introduce more vertices and edges or cause collapsing to take place .
however , in our case , we will see that each secondary relation is already satisfied at each vertex .
we chose this presentation of @xmath14 precisely for this reason .
the secondary relations , which must be true for all @xmath180 , are listed below .
we label each with the associated name of the primary relation preceded by `` s '' , for secondary .
the indices range over the same index sets as the corresponding primary relation .
@xmath316^{(q-1)/2}[12]^{|2w|}[13]^{(q+1)/2}}\\ s\delta ' : x&=x^{[31]^{2q}}\\ s\delta_i : x&=x^{[21]^i[13]^q[12]^{2i}[13]^q[21]^i}\\ s\theta_{ik } : x&=x^{[21]^{i+1}[13]^k12[13]^k[12]^{2i}[13]^k12[13]^k[21]^{i+1}}\\ s\kappa : x&=x^{[21]^{|4w|}}\\ s\iota_i : x&=x^{[12]^i13[21]^{2i}23[12]^{i+1}}\end{aligned}\ ] ] to examine the effect of the secondary relations at every element @xmath180 we compute the following displacements : @xmath317^{(q-1)/2}[12]^{|2w|}[13]^{(q+1)/2}&=-\langle 2|w| , 4q\rangle\\ \delta_{\bf01}{[31]^{2q}}&=\langle0 , 8q \rangle\\ \delta_{\bf01}{[21]^i[13]^q[12]^{2i}[13]^q[21]^i}&=(-1)^i\langle0 , -8q \rangle\\ \delta_{\bf01}[21]^{i+1}[13]^k12[13]^k[12]^{2i}[13]^k12[13]^k[21]^{i+1}&=\langle0,0 \rangle\\
\delta_{\bf01}[21]^{|4w|}&=\langle4|w|,0 \rangle\\ \delta_{\bf01}[12]^i13[21]^{2i}23[12]^{i+1}&=\langle0,0 \rangle\\ \ ] ] because horizontal translation by @xmath318 and vertical translation by @xmath319 preserve elements of the quandle in both @xmath156 and @xmath163 , it follows that the secondary relations hold at every vertex in @xmath154 and @xmath162 .
this completes the proof of theorem [ cayley graph ] in the case where @xmath92 is odd .
we leave the case when @xmath92 is even to the reader . in working through the details ,
notice that the construction of @xmath154 and @xmath153 run in parallel fashion and that the construction of @xmath162 is the same as in the case where @xmath92 is odd .
two - bridge links are not distinguished by their involutory quandles because @xmath320 , see @xcite . however , for @xmath321-montesinos links we have the following theorem . *
proof : * let @xmath74 and @xmath323 .
assume that @xmath324 , @xmath148 , and @xmath325 .
using corollary [ corollary ] to compute the number of components of the quandles , as well as their sizes we conclude that @xmath326 and @xmath327 .
if @xmath328 , then @xmath92 divides @xmath329 from which it follows that @xmath330 and @xmath331 .
hence the links are equivalent . on the other hand ,
if @xmath332 , then @xmath92 divides @xmath333 and we must have @xmath334 .
this gives that @xmath335 .
but these conditions now imply that @xmath25 and @xmath322 are mirror images of each other .
@xmath252 in any involutory quandle @xmath16 , the subquandle generated by two elements @xmath180 and @xmath47
is called the _ geodesic _ through @xmath180 and @xmath47 , which we denote @xmath336 .
a _ maximal _ geodesic is one that is not a proper subset of any other geodesic .
more details regarding geodesics are given in @xcite , @xcite .
the following theorem describes the set of maximal geodesics in any @xmath1-montesinos link . * if @xmath92 is odd , then one maximal geodesic is @xmath338 of length @xmath339 and the remaining @xmath92 maximal geodesics are @xmath340^{(q-1)/2}})$ ] , each of length @xmath341 .
the geodesic @xmath338 is equal to @xmath154 and the intersection of any two of the other geodesics is @xmath162 . * if @xmath92 is even , then one maximal geodesic is @xmath342 of length @xmath343 and the remaining @xmath92 maximal geodesics are @xmath344 , @xmath345 , @xmath346 , @xmath347 , @xmath348 , each of length @xmath341 .
the geodesic @xmath342 is equal to @xmath349 and the intersection of any two of the other geodesics is @xmath162 .
* proof : * suppose @xmath92 is odd and consider the geodesic generated by @xmath189 and @xmath350 . traveling only `` forward '' on this geodesic
, we generate the sequence @xmath351^k } , 1^{32[123132]^k } , \dots.\ ] ] note that @xmath352 and that @xmath353^k=-2k\langle 1 , 4 \rangle$ ] .
thus , the element @xmath354^k}$ ] is represented by the vertex @xmath355 in @xmath156 .
this is a vertex of type @xmath356 .
similarly , the element @xmath357^k}$ ] is represented by the vertex @xmath358 , a vertex of type * 13*. now consider moving any vertex @xmath359 of @xmath156 to an equivalent vertex which lies in the fundamental domain @xmath155 .
we can do this by first translating @xmath359 to @xmath360 , a vertex in the domain @xmath361 , by using translation in the horizontal direction by a multiple of @xmath362 and in the vertical direction by a multiple of @xmath319 .
if @xmath363 , then we may reflect @xmath360 through the point @xmath364 to obtain the point @xmath365 . if @xmath366 , then the element @xmath354^{q|w|}}$ ] located at @xmath367 is equivalent to the vertex @xmath368 which is the element @xmath189 .
similarly , the element @xmath357^{q|w|}}$ ] located at @xmath369 is equivalent to the vertex @xmath370 which is the element @xmath350 .
thus the geodesic has returned to the beginning after traversing @xmath343 points .
we want to show that all @xmath343 of these points are distinct in @xmath155 , and therefore make up all of the component @xmath154 . if not , then there are two distinct points @xmath359 and @xmath16 on the geodesic when thought of as points in @xmath156 that project to the same point in @xmath155 .
one possibility is that @xmath359 and @xmath16 translate to the same point @xmath371 in @xmath361 . because such a translation preserves the type of a vertex
, we see that @xmath359 and @xmath16 must both be of type * 01 * or * 13*. if this were true , there would exist integers @xmath372 and @xmath373 with @xmath374 such that @xmath375 and @xmath376 or @xmath377 and @xmath378 . in either case
, @xmath371 implies that @xmath379 ( mod @xmath380 ) and @xmath379 ( mod @xmath92 ) because @xmath92 is odd .
thus there exists @xmath6 and @xmath127 such that @xmath381 and @xmath382 .
furthermore , @xmath383 and @xmath384 . from this
we obtain that @xmath385 $ ] and so @xmath92 divides @xmath386 . because @xmath93 and @xmath92 are co - prime and @xmath387 , it follows that @xmath87 and @xmath388 .
another possibility is that @xmath389 but instead @xmath360 and @xmath390 are symmetric with respect to either the point @xmath159 or @xmath160 . but any pair of such symmetric points must have types @xmath356 and @xmath219 or types @xmath214 and @xmath391 .
this is impossible because @xmath359 and @xmath16 have types * 01 * or * 13 * and these types are preserved by the translations .
thus , the geodesic contains @xmath343 distinct points , filling out all of @xmath154 . a similar argument can be used to show that the geodesic generated by @xmath190 and @xmath270^k}$ ] , where @xmath392 alternates between @xmath162 and @xmath154 traversing every point of @xmath162 exactly once and every point of @xmath155 with @xmath47-coordinate equal to @xmath393 exactly once .
this is a total of @xmath318 points in @xmath154 and @xmath318 points in @xmath162 .
we leave this case , as well as the case of the geodesic generated by @xmath190 and @xmath270^k1}$ ] to the reader .
@xmath252 if @xmath16 is a quandle an _ automorphism _ of @xmath16 is a bijection @xmath394 such that @xmath395 for all @xmath180 and @xmath47 in @xmath16 .
the set of automorphism forms a group , @xmath396 , under composition in the obvious way .
any automorphism must take a maximal geodesic in an involutory quandle to a maximal geodesic .
using this idea we may prove the following theorem . *
proof : * first note that any automorphism must preserve both maximal geodesics and components . for simplicity , we consider the case where @xmath92 is odd .
the even case is similar .
if @xmath401 , there is one maximal geodesic of length @xmath339 which must be taken to itself and @xmath403 shorter geodesics of length @xmath341 that must be permuted among themselves .
since @xmath189 and @xmath350 generate the long geodesic , we must have that this pair of elements is sent to another pair of generators of this geodesic and knowing where they are sent is sufficient to determine the image of any other element of this geodesic . because the geodesic contains @xmath343 elements
, there are @xmath343 possible choices for @xmath404 . in order for @xmath404 and @xmath405 to generate the geodesic , we need these two points located a distance @xmath406 apart with @xmath406 prime to @xmath343 .
thus there are @xmath407 possibilities for @xmath405 , where @xmath402 is the euler @xmath402-function .
we also know that @xmath190 must be taken to another point of @xmath162 .
thus there are @xmath318 possibilities for its image . once this is determined , the image of every point is determined because each point lies on a maximal geodesic containing 2 .
thus there are at most @xmath408 elements in @xmath397 .
the proof for @xmath399 is similar . | in this paper we show that montesinos links of the form @xmath0 , which we call @xmath1-montesinos links , have finite involutory quandles .
this generalizes an observation of winker regarding the @xmath2-pretzel links .
we also describe some properties of these quandles . |
the observation that the low energy limit of certain open string theories can be interpreted as yang - mills theories living on a noncommutative ( nc ) manifold @xcite has raised renewed interest in noncommutative quantum field theories . furthermore , noncommutative quantum field theory ( ncqft ) in itself remains an appealing way to introduce a length scale and to cut off some short - distance contributions @xcite consistently with the symmetries of a given model . in the simplest case ,
the nc structure of spacetime is described by a set of _ constant c - number _ parameters @xmath3 , or , equivalently , by an energy scale @xmath4 and dimensionless parameters @xmath5 : @xmath6 = { \mathrm{i}}\theta^{\mu\nu } = { \mathrm{i}}\frac{1}{\lambda_{\text{nc}}^2 } c^{\mu\nu } = { \mathrm{i}}\frac{1}{\lambda_{\text{nc}}^2 } \begin{pmatrix } 0 & - e^1 & - e^2 & - e^3 \\
e^1 & 0 & - b^3 & b^2 \\
e^2 & b^3 & 0 & - b^1 \\
e^3 & - b^2 & b^1 & 0 \end{pmatrix}\,.\ ] ] here we have defined dimensionless `` electric '' and `` magnetic '' parameters @xmath7 and @xmath8 for convenience .
obviously , fixed non - vanishing values for @xmath3 result in a breaking of lorentz invariance .
a possible mechanism of spontaneous breaking is the condensation of a rank two antisymmetric background field . in this paper
, we will investigate phenomenological consequences of @xmath3 at hadron colliders and not deal with the dynamics of the symmetry breaking mechanism .
since the string scale is essentially unknown , we have also no prediction for @xmath4 .
the current best collider tests @xcite give @xmath9{gev}$ ] for a nc extension of qed , while experiments at a future photon collider are expected to reach @xcite .
tests of rotation invariance in atomic physics and astrophysical tests give substantially higher bounds , but assume nc geometry at large scales ( see @xcite for a review ) .
collider experiments do not probe space - time coordinates themselves , but functions of space - time coordinates , such as wave functions of external particles and effective lagrangians .
consequently , one can employ a deformed product to describe the scattering theory on a given nc manifold by an equivalent scattering theory on a commuting manifold with new effective interactions . in the case of the operator algebra ( [ eq : theta ] ) , one can use the moyal - weyl @xmath10-product @xcite @xmath11 to map products of functions of nc coordinates , @xmath12 , to the equivalent products of functions on ordinary space - time coordinates , @xmath13 .
it is easy to see that the commutation relation ( [ eq : theta ] ) is reproduced by the @xmath10-commutator : @xmath14(x ) = ( x^\mu \ast x^\nu)(x ) - ( x^\nu \ast x^\mu)(x ) = { \mathrm{i}}\theta^{\mu\nu}=[\hat{x}^\mu,\hat{x}^\nu]\,.\ ] ] the obvious approach to constructing ncqfts by replacing all field products by the @xmath10-product ( [ eq : moyal - weyl ] ) fails for general gauge theories .
for example , in such a nc version of qed all charged matter field must carry the same charge @xcite which makes this approach unsuitable for the @xmath15 hypercharge of the standard model .
furthermore , this approach does not work for @xmath16 gauge theories , because the @xmath10-commutator of two infinitesimal gauge transformations does not close in the @xmath16 lie algebra @xcite : @xmath17 = [ \alpha^at^a \operatorname{\stackrel{\ast}{,}}\beta^bt^b ] = \frac{1}{2}\{\alpha^a\operatorname{\stackrel{\ast}{,}}\beta^b\}[t^a , t^b ] + \frac{1}{2}[\alpha^a\operatorname{\stackrel{\ast}{,}}\beta^b]\{t^a , t^b\ } \,.\ ] ] in order to circumvent these problems , we follow the approach of @xcite and let the nc gauge field @xmath18 and gauge parameter @xmath19 take their values in the enveloping algebra of the lie algebra of the standard model ( sm ) gauge group @xmath20 .
@xmath18 , @xmath19 and the nc matter fields @xmath21 are expressed as nonlinear functions of ordinary fields @xmath22 , @xmath23 and @xmath24 via seiberg - witten maps ( swm ) @xcite such that the nc gauge transformations of the former , [ eq : nc - gauge - transformations ] @xmath25 + \frac{({\mathrm{i}}g)^2}{2 ! } [ \hat\lambda\operatorname{\stackrel{\ast}{,}}[\hat\lambda\operatorname{\stackrel{\ast}{,}}\hat a_\mu ] ] + \partial_\mu \hat\lambda + { \mathrm{i}}g [ \hat\lambda\operatorname{\stackrel{\ast}{,}}\partial_\mu \hat\lambda ] + \mathcal{o}(\hat\lambda^3)\ , , \notag\end{aligned}\ ] ] are realized by ordinary gauge transformations of the latter : [ eq : gauge - transformations ] @xmath26 + \frac{({\mathrm{i}}g)^2}{2 ! } [ \lambda , [ \lambda , a_\mu ] ] + \partial_\mu \lambda + { \mathrm{i}}g
[ \lambda,\partial_\mu \lambda ] + \mathcal{o}(\lambda^3)\ , .
\notag\end{aligned}\ ] ] this leads to the so - called gauge - equivalence conditions [ eq : swm - condition ] @xmath27 that can be solved order by order in @xmath3 .
the first order is given by @xmath28 \psi(x ) + \mathcal{o}(\theta^2)\\ \hat \lambda(x ) & = \lambda(x ) + \frac{1}{4 } \theta^{\rho\sigma } \left\{a_\sigma(x ) , \partial_\rho \lambda(x)\right\ } + \mathcal{o}(\theta^2)\end{aligned}\ ] ] with @xmath29 $ ] .
the swm approach also solves the charge quantization problem by introducing a separate nc gauge field for each eigenvalue of the @xmath15 charge operator .
each of these gauge fields is subsequently mapped by the swm to the _ same _ sm @xmath15 field .
thus no additional degrees of freedom appear .
this approach has been used to construct a nc generalization of the complete electroweak standard model @xcite , which will be referred to as ncsm below .
this construction leads to modifications of the familiar sm interaction vertices as well as to new interactions that are forbidden in the sm .
the most striking feature of the ncsm is that it allows direct interactions among neutral gauge bosons , depending on the representation of the enveloping algebra chosen for the trace in the gauge boson lagrangian @xcite @xmath30 note that all representations are equivalent up to a renormalization only in the case of lie algebras , but not in the corresponding enveloping associative algebra .
we should point out that nc gauge theories constructed from commutative gauge theories using the swm are not renormalizable by field and gauge coupling renormalization alone , since the @xmath31-point functions of matter fields require additional counterterms that can not be obtained from the application of the swm @xcite .
nevertheless , the ncsm can be employed as a well defined effective field theory , because any chiral model that is constructed from an anomaly free chiral model using the swm has been shown to be anomaly free @xcite .
to first order in @xmath3 , all modifications of the sm correspond to effective dimension - six operators . the latter lead to observable deviations from the sm predictions or , in the absence of deviations , to bounds on @xmath4 .
initially , most phenomenological studies have focused on rare and @xmath32-decays @xcite , while tests at future colliders have been considered in @xcite .
additional phenomenological results have been presented in @xcite . in this paper
, we report on a phenomenological study of the ncsm in first order in @xmath3 , concentrating on the process @xmath33 at hadron colliders .
we shall point out experimental signatures unique to the ncsm and estimate the range of the nc scale @xmath4 that can be probed at the tevatron and the lhc .
the paper is structured as follows : in section [ sec : xsect ] , we discuss the scattering amplitude that gives rise to the signals described in section [ sec : signals ] . in section [ sec : analysis ] , we analyse the results of our monte carlo ( mc ) simulations and estimate the discovery reach . section [ sec : concl ] gives a short outlook on future work .
feynman rules and analytical results for the cross section are presented in the appendix .
the pair production of neutral electroweak gauge bosons @xmath34 at a hadron collider is described by the partonic process @xmath35 .
the corresponding squared matrix element at @xmath36 is given by @xmath37 where @xmath38 is the sm amplitude with @xmath39 and @xmath40 denotes the new contributions arising in the ncsm .
the relevant feynman rules are provided in appendix [ app : feynman ] .
below , the @xmath36 contributions to the sm vertices ( [ eq : vertex - ffv ] ) have been marked by an open box in the corresponding sm feynman diagrams : [ eq : nc - amplitude ] @xmath41 the contact term @xmath42 with the vertex given in ( [ eq : vertex - ffvv ] ) , is required by gauge invariance , as can be seen from the ward identity ( [ eq : wia ] ) .
in contrast , the two @xmath43-channel diagrams @xmath44 involving the triple gauge boson couplings given in ( [ eq : vertex - vvv ] ) are separately gauge invariant .
the corresponding ward identities @xmath45 have been proven in @xcite .
the separate gauge invariance of @xmath46 and @xmath47 is not surprising , because the strength of the triple gauge boson interactions depends on the choice of representation of the enveloping algebra @xcite used for the gauge field lagrangian ( [ eq : tr(f^2 ) ] ) .
the new coupling constants @xmath48 and @xmath49 are only constrained by the matching to the sm in the limit @xmath50 and can vary independently in a finite range @xcite , as illustrated in fig .
[ fig : k_zgg / k_zzg ] .
range for the triple neutral gauge boson couplings @xmath48 and @xmath49 allowed by the matching to the sm in the limit @xmath50 in the nonminimal ncsm , as defined in @xcite . ]
it has been shown @xcite for @xmath51-collisions that polarization is essential for obtaining a non - vanishing signal in @xmath36 . at the lhc
one will not have polarized initial states , neither can one expect to observe the polarization of outgoing photons or use @xmath52-decays as a polarimeter .
therefore , we have concentrated on unpolarized @xmath2-production . the analytical result for @xmath53
is presented in appendix [ app : xsect ] . without polarization ,
the most important observable , i.e. an azimuthal dependence of the cross section , vanishes for symmetric initial and final states , as we will see in section [ sec : signals ] . in the present case ,
it is the axial coupling of the @xmath52 that will engender an observable deviation from the sm prediction .
since the width of the intermediate @xmath52 is negligible for @xmath54 , one needs another source for an imaginary part in the scattering amplitudes to compensate the factor of @xmath55 that comes with @xmath3 and to obtain a non - vanishing contribution to the @xmath36 interference . in polarized scattering
, the polarization vectors and spinors can provide this imaginary part , but all the polarization sums are real , of course .
however , the @xmath56 factor in the axial coupling of the @xmath52 produces a factor @xmath57 after tracing and thus creates a non - vanishing effect from contractions with the three independent momenta and @xmath3 .
the relevant hadronic processes are @xmath59 at the lhc and @xmath60 at the tevatron with subsequent decays of the @xmath52 into @xmath61 and @xmath62 . for these
, the cross sections and lepton distributions have been calculated using the results for the hard @xmath63 cross sections from section [ sec : xsect ] together with the phase space generation and parton distribution functions provided by whizard @xcite for monte carlo simulation .
as expected for new short distance physics , we observe a deviation of the photon @xmath64 distribution at high @xmath64 from the sm prediction .
however , this deviation is not specific to the ncsm and , in addition , turns out to be smaller than a few @xmath65 for @xmath66 . on the other hand , the nc parameter @xmath3 breaks rotational invariance and leads to a characteristic dependence of the cross sections on the azimuthal angle @xmath67 , which can be used to discriminate against other new physics effects .
for this reason , we have focussed our phenomenological analysis on the azimuthal distribution of the photon .
as has been shown previously @xcite , the @xmath68 amplitude in the ncsm depends only on @xmath69 and @xmath70 .
in contrast , the @xmath71 amplitude also depends on @xmath72 and @xmath73 , due to the axial @xmath74 couplings .
in addition , a dependence on @xmath75 appears because the vector bosons are not aligned with the beam axis @xmath76 .
nevertheless , the cms parton cross sections show a much stronger dependence on the components @xmath7 than on @xmath8 , as exemplified in fig .
[ fig : azimuthal / e / b ] . to be precise , this is true everywhere except sufficiently close to the polar angle @xmath77 , where the dependence on @xmath7 vanishes due to the antisymmetry of the @xmath36 contribution to @xmath78 in @xmath79 .
azimuthal distribution of the @xmath80 in @xmath81 at fixed cms energy @xmath82 for @xmath83 and @xmath84 , in comparison with @xmath85 for @xmath86 .
] however , at hadron colliders , the partonic cms of most events is boosted significantly along the beam axis . as is well known from electrodynamics , @xmath7 and @xmath8 are mixed by lorentz boosts along the beam axis @xmath76 : @xmath87 ( @xmath88 , @xmath89 ) .
the measurements of @xmath90 and @xmath91 , respectively , are therefore highly correlated by kinematics .
we have verified that the strenght of the correlation is essentially determined by the expectation value of the boost @xmath92 .
[ sec : unphysical ] calculating the scattering amplitude only to first order in the noncommutativity @xmath3 , one must neglect the @xmath93 contributions in the squared amplitude for consistency .
therefore , in this approximation , the deviations from the sm only come from the interference of @xmath94 with @xmath95 in ( [ eq:|a|^2 ] ) . as a consequence
, the @xmath36 cross section can even become negative in some regions of phase space for sufficiently large absolute values and appropriate signs of components of @xmath3 , such that the nc / sm interference dominates over the sm contribution .
the relative size of the interference term is determined by @xmath96 , @xmath82 being the partonic cms energy .
there is a wide range of possible values of @xmath82 in high energy hadron collisions , but the most statistics will be collected at moderate values of @xmath82 .
thus in @xmath36 , values of @xmath4 that cause observable deviations at such moderate values of @xmath82 can lead to unphysical cross sections at the highest @xmath82 available .
this problem is not specific to simulations in the ncsm , but common to all studies of new physics that can be parametrized by anomalous couplings .
a pragmatic solution is to unitarize the contributions from new physics by applying appropriate form factors that cut off the unphysical effects .
since there are very few events to be expected at the highest cms energies , the conclusions should not depend on the details of the form factors .
therefore , we have simply replaced @xmath78 by @xmath97 everywhere in our simulations . this solution is also legitimized by the expectation that higher orders in @xmath3 will damp the large negative interference contributions .
in fact , preliminary results to second order in @xmath3 support this expectation @xcite .
in order to obtain a realistic estimate of the sensitivity at the tevatron and the lhc , one has to take into account backgrounds , detector effects and selection cuts .
clearly , a comprehensive analysis of all reducible backgrounds and detector effects is beyond the scope of a theoretical study and must eventually be performed by the experimental collaborations .
however , the final states under consideration are simple enough for a phenomenological analysis based on simple cuts .
moreover , experience at the tevatron @xcite indicates that the combined detection efficiency for @xmath98 can be assumed to be larger than @xmath99 .
all numerical results presented below are obtained for a @xmath100 efficiency .
smaller uniform efficiencies can easily be taken into account by scaling up the integrated luminosity accordingly .
we have simulated the process @xmath101 at @xmath102{tev}$ ] using the program package whizard @xcite and demanding the following acceptance cuts on leptons and photons : [ eq : acceptance ] @xmath103{gev } \\
\theta(\ell^\pm ) , \theta(\gamma ) & \ge 5^\circ \\
\theta(\ell^\pm ) , \theta(\gamma ) & \le 175^\circ \\ p_t(\ell^\pm ) , p_t(\gamma ) & \ge \unit[10]{gev}\,.\end{aligned}\ ] ] for the tevatron , we have simulated @xmath104 at @xmath105{tev}$ ] using the same acceptance cuts ( [ eq : acceptance ] ) . for simplicity , we have assumed that the components of @xmath3 remain aligned with the beam axis and the detector over the time of the measurement .
this assumption is not justified , because we should expect that @xmath3 is aligned with a fixed cosmic reference frame , that is determined by the dynamics of the underlying string theory .
therefore the alignment of the detector must be recorded with each event and the combined effect of the earth s rotation and revolution must be taken into account .
this poses no principal difficulty and will not change our conclusions .
the hard scattering process @xmath81 ( [ eq : nc - amplitude ] ) with subsequent decays @xmath106 has the drell - yan process @xmath107 with photon radiation and @xmath108 with @xmath109 as irreducible backgrounds .
both have been taken into account in our calculation . as explained in section [ sec : xsect ] , the unpolarized azimuthal distribution in @xmath110 is flat .
we suppress this background by requiring @xmath111 in order to also reduce the radiative drell - yan events , we have applied an angular separation cut of @xmath112 ( cf .
@xcite ) . finally , we require a minimum and maximum total energy in the partonic cms : @xmath113{gev } \leq |m(\ell^+\ell^-\gamma)| \leq \unit[1]{tev}\,.\ ] ] the lower cut enriches the signal , while the upper cut reduces the influence of the elimination of the unphysical parton cross sections discussed in section [ sec : unphysical ] . at the lhc , @xmath114- and @xmath115-collisions
will occur with identical rates . as shown in fig .
[ fig : azimuthal / e / b ] , the dominant @xmath69 and @xmath70 contributions to the deviation of the parton cross sections from the sm prediction is antisymmetric in @xmath116 and will therefore cancel for @xmath117 initial states unless additional cuts are applied that separate events originating from @xmath118 and @xmath119 . in the proton ,
the average momentum fraction of the light valence quarks is much higher than that of the anti - quarks which exist only in the sea . as a result
, all @xmath114-events will be boosted strongly in the direction of the quark .
therefore we can enrich our samples of signal events by requiring a minimal boost in the appropriate direction .
we have found that demanding the momenta of both the photon and the lepton pair to lie in the _ same _ hemisphere in the laboratory frame , that is @xmath120 produces a nice signal , as displayed in fig .
[ fig : azimuthal ] . in the figure , we have chosen a rather low value of @xmath121 for illustration .
the values of @xmath48 and @xmath49 correspond to the top corner of the polygon in fig .
[ fig : k_zgg / k_zzg ] . in the case of the tevatron ,
the rle of quarks and anti - quarks are reversed in the antiproton .
therefore , we demand in this case that the momenta of the photon and the lepton pair lie in _ opposite _ hemispheres . for sufficiently small @xmath4 ,
the azimuthal distribution of the @xmath80 is similar to the distribution plotted in fig .
[ fig : azimuthal ] .
azimuthal distributions of the @xmath80 in @xmath122 at the lhc , @xmath123 , @xmath124 , for @xmath125 , @xmath126 , @xmath127 in the sm ( open ) and the ncsm ( full ) with @xmath121 , @xmath128 , @xmath129 , @xmath130 . ] in the following , we analyze the nc effects on the azimuthal dependence of the unpolarized cross section stemming from the components @xmath131 of @xmath1 .
we disregard the small contribution from @xmath75 , since , by symmetry , it has no influence on the azimuthal distribution . furthermore , we have taken advantage of the fact that the deviation of the cross section from the sm prediction is linear in @xmath1 so that the corresponding binned @xmath132-function @xmath133 depends quadratically on @xmath1 : @xmath134 up to statistical fluctuations .
the elements of the matrix @xmath135 in ( [ eq : chi^2/x^2 ] ) are then determined from a fit of @xmath136 to the right - hand side of ( [ eq : chi^2 ] ) for a sufficiently large grid of values @xmath137 .
subsequently , we can diagonalize @xmath135 and obtain the error ellipses that describe the experimental reach of collider experiments . the @xmath138 ( dark ) and @xmath139 ( light )
exclusion contours for @xmath140{gev}$ ] and @xmath141{fb^{-1}}$ ] at the lhc discussed in the text.,title="fig:",scaledwidth=30.0% ] the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath140{gev}$ ] and @xmath141{fb^{-1}}$ ] at the lhc discussed in the text.,title="fig:",scaledwidth=30.0% ] the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath140{gev}$ ] and @xmath141{fb^{-1}}$ ] at the lhc discussed in the text.,title="fig:",scaledwidth=30.0% ] + the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath140{gev}$ ] and @xmath141{fb^{-1}}$ ] at the lhc discussed in the text.,title="fig:",scaledwidth=30.0% ] the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath140{gev}$ ] and @xmath141{fb^{-1}}$ ] at the lhc discussed in the text.,title="fig:",scaledwidth=30.0% ] the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath140{gev}$ ] and @xmath141{fb^{-1}}$ ] at the lhc discussed in the text.,title="fig:",scaledwidth=30.0% ] for the lhc , the results of the likelihood fits are shown in fig . [
fig : contours ] , setting @xmath140{gev}$ ] and @xmath142 . from the error
ellipses in the first and second column of fig .
[ fig : contours ] , we derive the following sensitivity limit on the nc scale for an integrated luminosity of @xmath143 ( with @xmath100 detection efficiency ) : @xmath144{tev } \;\text{for}\ ; |\vec e|^2 + |\vec b|^2 = 1\,.\ ] ] in the presence of triple gauge couplings among neutral gauge bosons , these limits do not change significantly .
this analysis indicates that the current collider bounds @xcite can be improved at the lhc by an order of magnitude .
the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath145{gev}$ ] and @xmath146{fb^{-1}}$ ] at the tevatron discussed in the text.,title="fig:",scaledwidth=30.0% ] the @xmath138 ( dark ) and @xmath139 ( light ) exclusion contours for @xmath145{gev}$ ] and @xmath146{fb^{-1}}$ ] at the tevatron discussed in the text.,title="fig:",scaledwidth=30.0% ] a similar analysis of the likelihood fits for the tevatron , shown in fig .
[ fig : contours - tevatron ] , for @xmath147{fb^{-1}}$ ] implies that the sensitivity on the nc scale reaches @xmath148{gev}$ ] , which is comparable to existing lep bounds @xcite .
as expected from the lorentz boost @xmath149 , we find that in the laboratory frame measurements of @xmath69 are correlated with @xmath73 and measurements of @xmath70 with @xmath72 .
the corresponding @xmath138 and @xmath139 contours are depicted for @xmath140{gev}$ ] in the right column of fig .
[ fig : contours ] .
since the dependence on @xmath72 and @xmath73 in the partonic cms is very weak , one expects very elongated ellipses in the laboratory frame , in agreement with our result . due to statistical fluctuations
, the fitted matrix @xmath135 can have a negative eigenvalue , as happened in the bottom right plot of fig .
[ fig : contours ] .
this sign is unphysical and as expected changes with the random number sequence used in the simulations . for all practical purposes , the error ellipses for @xmath90 and @xmath91
should be viewed as straight lines . in the other pairs
@xmath150 , @xmath151 and @xmath152 , we find no correlations . indeed , the only source of violation of rotational invariance is @xmath3 itself . therefore no correlations between @xmath69 and @xmath70 , as well as between @xmath72 and @xmath73 , are expected since the measurements are related by a rotation around the beam axis by @xmath153 .
having established the absence of correlations that are not of purely kinematical origin , one can avoid expensive non - linear @xmath31-parameter fits in subsequent work that will take higher orders in @xmath3 into account @xcite .
we have studied the effect of a noncommutative extension of the sm on the production of neutral vector bosons at the lhc and the tevatron .
we have shown that , under conservative assumptions , values for the nc scale of @xmath154{tev}$ ] can be probed at the lhc , while the tevatron is able to confirm existing limits .
the analysis has been performed to first order in @xmath3 . in this approximation
, unphysical cross sections appear for large @xmath82 and for @xmath4 in the region of sensitivity .
these have been handled by a pragmatic regularization described in section [ sec : signals ] and appropriate kinematical cuts .
on general grounds , one expects the problem to disappear when higher orders in @xmath3 are taken into account .
preliminary results for the second order contributions confirm this expectation .
we also find that cancellations in the unpolarized cross sections at @xmath36 are not effective in higher orders , allowing also studies of processes with symmetric final states like @xmath155 , one of the most important channels at the lhc .
therefore , the derivation of seiberg - witten maps for @xmath20 of the corresponding ncsm in higher orders is important for a comprehensive analysis of the ncsm at the lhc
. the construction of the ncsm in @xmath93 and further phenomenological tests , including @xmath51 final states , will be presented in future publications @xcite .
this research is supported by deutsche forschungsgemeinschaft , grant ru311/1 - 1 , research training group 1147 _ theoretical astrophysics and particle physics _ and by bundesministerium fr bildung und forschung germany , grant 05ht4wwa/2 .
a.a . gratefully acknowledges support from evangelisches studienwerk e.v .
villigst . 99 n. seiberg , e. witten , jhep * 09 * , 32 ( 1999 ) [ arxiv : hep - th/9908142 ] .
h. s. snyder , phys .
71 * , 38 ( 1947 ) .
g. abbiendi _ et al . _
[ opal collaboration ] , phys . lett . *
b568 * , 181 ( 2003 ) [ arxiv : hep - ex/0303035 ] .
t. ohl and j. reuter , phys .
* d70 * , 076007 ( 2004 ) [ arxiv : hep - ph/0406098 ] ; t. ohl and j. reuter [ arxiv : hep - ph/0407337 ] .
i. hinchliffe , n. kersting and y. l. ma , int . j. mod
a19 * , 179 ( 2004 ) [ arxiv : hep - ph/0205040 ] .
j. e. moyal , proc .
* 45 * , 99 ( 1949 ) .
m. hayakawa , phys .
lett . , * b478 * , 394 ( 2000 ) [ arxiv : hep - th/9912094 ] .
j. madore , s. schraml , p. schupp and j. wess , eur .
j. * c16 * , 161 ( 2000 ) [ arxiv : hep - th/0001203 ] ; b. jurco , s. schraml , p. schupp and j. wess , eur .
j. * c17 * , 521 ( 2000 ) [ arxiv : hep - th/0006246 ] ; j. wess , commun .
. phys . * 219 * 247 ( 2001 ) ; b. jurco , l. mller , s. schraml , p. schupp and j. wess , eur .
j. * c21 * , 383 ( 2001 ) [ arxiv : hep - th/0104153 ] .
x. calmet , b. juro , p. schupp , j. wess , m. wohlgenannt , eur .
j. * c23 * , 363 ( 2002 ) [ arxiv : hep - ph/0111115 ] .
r. wulkenhaar , jhep * 0203 * , 024 ( 2002 ) [ arxiv : hep - th/0112248 ] ; m. buric and v. radovanovic , class .
* 22 * , 525 ( 2005 ) [ arxiv : hep - th/0410085 ] ; m. buric , d. latas and v. radovanovic , jhep * 0602 * , 046 ( 2006 ) [ arxiv : hep - th/0510133 ] . c. p. martin , nucl .
b * 652 * , 72 ( 2003 ) [ arxiv : hep - th/0211164 ] ; f. brandt , c. p. martin and f. r. ruiz , jhep * 0307 * , 068 ( 2003 ) [ arxiv : hep - th/0307292 ] .
w. behr , n. g. deshpande , g. duplancic , p. schupp , j. trampetic and j. wess , eur .
j. * c29 * , 441 ( 2003 ) [ arxiv : hep - ph/0202121 ] ; p. schupp , j. trampetic , j. wess and g. raffelt , eur .
j. * c36 * , 405 ( 2004 ) [ arxiv : hep - ph/0212292 ] ; p. minkowski , p. schupp and j. trampetic , eur .
j. * c37 * , 123 ( 2004 ) [ arxiv : hep - th/0302175 ] ; b. melic , k. passek - kumericki and j. trampetic , phys . rev . *
d72 * , 054004 ( 2005 ) [ arxiv : hep - ph/0503133 ] ; b. melic , k. passek - kumericki and j. trampetic , phys . rev . *
d72 * , 057502 ( 2005 ) [ arxiv : hep - ph/0507231 ] . c. e. carlson , c. d. carone and r. f. lebed , phys
b518 * , 201 ( 2001 ) [ arxiv : hep - ph/0107291 ] ; m. haghighat , m. m. ettefaghi and m. zeinali , phys . rev . *
d73 * , 013007 ( 2006 ) [ arxiv : hep - ph/0511042 ] ; c. p.
martin and c. tamarit , jhep * 0602 * , 066 ( 2006 ) [ arxiv : hep - th/0512016 ] ; m. mohammadi najafabadi , phys
. rev . * d74 * , 025021 ( 2006 ) [ arxiv : hep - ph/0606017 ] .
w. kilian , _ whizard 1.0 : a generic monte - carlo integration and event generation package for multi - particle processes .
, lc - tool-2001 - 039 , ` http://www-ttp.physik.uni-karlsruhe.de/whizard/ ` ; w. kilian , _ whizard : complete simulations for electroweak multi - particle processes _ , in _ proceedings of ichep , amsterdam 2002 , p. 831_.
a. alboteanu , r. ohl , r. rckl ( in preparation ) .
f. abe _ et al . _
[ cdf collaboration ] , phys .
* 74 * , 1941 ( 1995 ) ; d. acosta _ et al . _
[ cdf ii collaboration ] , phys .
lett . * 94 * , 041803 ( 2005 ) [ arxiv : hep - ex/0410008 ] .
s. abachi _ et al . _
[ d0 collaboration ] , phys .
lett . * 75 * , 1028 ( 1995 ) [ arxiv : hep - ex/9503010 ] ; s. abachi _ et al . _
[ d0 collaboration ] , phys .
lett . * 78 * , 3640 ( 1997 ) [ arxiv : hep - ex/9702011 ] ; b. abbott _ et al . _ [ d0 collaboration ] , phys . rev . *
d57 * , 3817 ( 1998 ) [ arxiv : hep - ex/9710031 ] ; v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 95 * , 051802 ( 2005 ) [ arxiv : hep - ex/0502036 ] .
b. melic , k. passek - kumericki , j. trampetic , p. schupp and m. wohlgenannt , eur .
j. * c42 * , 483 ( 2005 ) [ arxiv : hep - ph/0502249 ] ; b. melic , k. passek - kumericki , j. trampetic , p. schupp and m. wohlgenannt , eur .
j. * c42 * , 499 ( 2005 ) [ arxiv : hep - ph/0503064 ] .
t. ohl , _ omega : an optimizing matrix element generator _ , in _ proceedings of 7th international workshop on advanced computing and analysis techniques in physics research ( acat 2000 ) _
( fermilab , batavia , il , 2000 ) [ arxiv : hep - ph/0011243 ] ; m. moretti , t. ohl , j. reuter , [ arxiv : hep - ph/0102195 ] ; t. ohl , j. reuter , c. schwinn ,
( long write - up , in preparation ) .
the feynman rules corresponding to the ncsm lagrangian @xcite have been derived in @xcite . here
, we collect only the neutral current couplings entering our analysis . choosing all momenta as incoming and using the shorthand notation @xmath156 and @xmath157 , one has @xmath158 in the above , the sm vector and axial - vector couplings in the fermion vertices
are given by @xmath159 while in the @xmath160 and @xmath161 vertices @xmath162 + \left ( k_1 \theta k_2 \right ) \left [ k_{3,\mu_1 } \eta_{\mu_2\mu_3 } - \eta_{\mu_1\mu_3 } k_{3,\mu_2 } \right ] \\
+ \bigl [ ( k_1\theta)_{\mu_1 } \left [ k_{2,\mu_3 } k_{3,\mu_2 } - ( k_2 k_3 ) g_{\mu_2\mu_3 } \right ] - ( \mu_1 \leftrightarrow \mu_2 ) - ( \mu_1 \leftrightarrow \mu_3 ) \bigr ] \\ + \text{cyclical permutations of}\ ; \bigl\ { ( \mu_1,k_1 ) , ( \mu_2,k_2 ) , ( \mu_3,k_3 ) \ , \bigr\}\,.\end{gathered}\ ] ] in the minimal ncsm , the coupling constants @xmath48 and @xmath49 vanish . in the non - minimal ncsm
, they can take the values shown in fig .
[ fig : k_zgg / k_zzg ] ( see @xcite ) .
in order to be able to express the partonic cross section for @xmath163 compactly , we introduce some further abbreviations . for contractions with the totally antisymmetric @xmath164 tensor we use the notation note that the @xmath43-channel contributions differ only by the coupling constants and the propagator because the triple gauge couplings contribute the same factor @xmath172\,.\end{gathered}\ ] ]
this analytical result has been verified numerically by comparing with the scattering amplitudes calculated with the help of an unpublished extension of the matrix element generator omega
the latter has also been employed for the calculation of the complete @xmath173 scattering amplitude used in the simulations . | we study collider signals for the noncommutative extension of the standard model using the seiberg - witten maps for @xmath0 to first order in the noncommutativity parameters @xmath1 . in particular , we investigate the sensitivity of @xmath2-production at the tevatron and the lhc to the components of @xmath1 .
we discuss the range of validity of this approximation and estimate exclusion limits from a monte carlo simulation . |
in this section we introduce the mathematical model describing incompressible isothermal flow in porous medium without reaction .
the considered equations for the velocity and pressure fields are for flows in fluid saturated porous media .
most of research results for flows in porous media are based on the darcy equation which is considered to be a suitable model at a small range of reynolds numbers . however , there are restrictions of darcy equation for modeling some porous medium flows , e.g. in closely packed medium , saturated fluid flows at slow velocity but with relatively large reynolds numbers .
the flows in such closely packed medium behave nonlinearly and can not be modelled accurately by the darcy equation which is linear .
the deficiency can be circumvented with the brinkman forchheimer - extended darcy law for flows in closely packed media , which leads to the following model : let @xmath0 , @xmath1 , represent the reactor channel .
we denote its boundary by @xmath2 .
the conservation of volume - averaged values of momentum and mass in the packed reactor reads as follows @xmath3 where @xmath4 , @xmath5 denote the unknown velocity and pressure , respectively . the positive quantity @xmath6 stands for porosity which describes the proportion of the non - solid volume to the total volume of material and varies spatially in general .
the expression @xmath7 represents the friction forces caused by the packing and will be specified later on .
the right - hand side @xmath8 represents an outer force ( e.g. gravitation ) , @xmath9 the constant fluid density and @xmath10 the constant kinematic viscosity of the fluid , respectively .
the expression @xmath11 symbolizes the dyadic product of @xmath12 with itself .
the formula given by ergun @xcite will be used to model the influence of the packing on the flow inertia effects @xmath13 thereby @xmath14 stands for the diameter of pellets and @xmath15 denotes the euclidean vector norm .
the linear term in ( [ s2eq2 ] ) accounts for the head loss according to darcy and the quadratic term according to forchheimer law , respectively . for the derivation of the equations , modelling and homogenization questions in porous media we refer to e.g. @xcite . to close the system ( [ s2eq1 ] )
we prescribe dirichlet boundary condition @xmath16 whereby @xmath17 has to be fulfilled on each connected component @xmath18 of the boundary @xmath19 .
we remark that in the case of polygonally bounded domain the outer normal vector @xmath20 has jumps and thus the above integral should be replaced by a sum of integrals over each side of @xmath19 .
the distribution of porosity @xmath21 is assumed to satisfy the following bounds @xmath22 with some constants @xmath23 .
a comprehemsive account of fluid flows through porous media beyond the darcy law s valid regimes and classified by the reynolds number , can be found in , e.g. , @xcite . also , see @xcite for simulating pumped water levels in abstraction boreholes using such nonlinear darcy - forchheimer law , and @xcite , @xcite , and @xcite for recent referenes on this model . in the next section we use the porosity distribution which is estimated for packed beds consisting of spherical particles and
takes the near wall channelling effect into account .
this kind of porosity distribution obeys assumption .
let us introduce dimensionless quantities @xmath24 whereby @xmath25 denotes the magnitude of some reference velocity .
for simplicity of notation we omit the asterisks .
then , the reactor flow problem reads in dimensionless form as follows @xmath26 where @xmath27 with @xmath28 and the reynolds number is defined by @xmath29 the existence and uniqueness of solution of the nonlinear model ( [ s2eq4 ] ) with constant porosity and without the convective term has been established in @xcite .
we will extend this result to the case when the porosity depends on the location and with the convective term in this work .
[ remarknse_1 ] becomes a navier - stokes problem if @xmath30 .
* notation * throughout the work we use the following notations for function spaces . for @xmath31 , @xmath32 and bounded subdomain @xmath33 let @xmath34 be the usual sobolev space equipped with norm @xmath35 .
if @xmath36 , we denote the sobolev space by @xmath37 and use the standard abbreviations @xmath38 and @xmath39 for the norm and seminorm , respectively .
we denote by @xmath40 the space of @xmath41 functions with compact support contained in @xmath42 .
furthermore , @xmath43 stands for the closure of @xmath40 with respect to the norm @xmath38 .
the counterparts spaces consisting of vector valued functions will be denoted by bold faced symbols like @xmath44^n$ ] or @xmath45^n$ ] .
the @xmath46 inner product over @xmath33 and @xmath47 will be denoted by @xmath48 and @xmath49 , respectively . in the case
@xmath50 the domain index will be omitted . in the following
we denote by @xmath51 the generic constant which is usually independent of the model parameters , otherwise dependences will be indicated .
in the following the porosity @xmath21 is assumed to belong to @xmath52 .
we start with the weak formulation of problem and look for its solution in suitable sobolev spaces .
let @xmath53 be the space consisting of @xmath46 functions with zero mean value .
we define the spaces @xmath54 and @xmath55 let us introduce the following bilinear forms @xmath56 & b:\,\boldsymbol{x}\times q&\to{{\mathbb r}}\,,&&\qquad b(\boldsymbol u , q)&=\bigl(\textrm{div}(\varepsilon\boldsymbol{u}),q\bigr)\,,\\[1.5ex ] & c:\,\boldsymbol{x}\times \boldsymbol{x}&\to{{\mathbb r}}\,,&&\qquad c(\boldsymbol u,\boldsymbol v)&=\frac{1}{re}\bigl(\alpha\boldsymbol{u},\boldsymbol{v}\bigr)\,.\\ \end{alignedat}\ ] ] furthermore , we define the semilinear form @xmath57 and trilinear form @xmath58 we set @xmath59 multiplying momentum and mass balances in ( [ s2eq4 ] ) by test functions @xmath60 and @xmath61 , respectively , and integrating by parts implies the weak formulation : + find @xmath62 with @xmath63such that @xmath64 first , we recall the following result from @xcite : [ s3thm1 ] the mapping @xmath65 is an isomorphism from @xmath66 onto itself and from @xmath67 onto itself .
it holds for all @xmath68 @xmath69 in the following the closed subspace of @xmath70 defined by @xmath71 will be employed .
next , we establish and prove some properties of trilinear form @xmath72 and nonlinear form @xmath73 .
[ s3lem2 ] let @xmath74 and @xmath75 with @xmath76 and @xmath77 .
then we have @xmath78 furthermore , the trilinear form @xmath72 and the nonlinear form @xmath73 are continuous , i.e. @xmath79 and for a sequence @xmath80 with @xmath81 , we have also @xmath82 we follow the proof of ( * ? ? ?
* lemma 2.1 , 2 , chapter iv ) and adapt it to the trilinear form @xmath83 which has the weighting factor @xmath21 . hereby , symbols with subscripts denote components of bold faced vectors , e.g. @xmath84 .
let @xmath85 , @xmath86 and @xmath87 . integrating by parts and employing density argument , we obtain immediately @xmath88 from sobolev embedding @xmath89 ( see @xcite ) and hlder inequality follows @xmath90 and consequently the proof of is completed . since @xmath91 and @xmath92 , the continuity estimate implies @xmath93 the continuity of @xmath73 follows from hlder inequality and sobolev embedding @xmath89 ( see @xcite ) @xmath94 in the next stage we consider the difficulties caused by prescribing the inhomogeneous dirichlet boundary condition .
analogous difficulties are already encountered in the analysis of navier stokes problem .
we will carry out the study of three dimensional case .
the extension in two dimensions can be constructed analogously . since @xmath95
, we can extend @xmath96 inside of @xmath97 in the form of @xmath98 with some @xmath99 .
the operator @xmath100 is defined then as @xmath101 we note that in the two dimensional case the vector potential @xmath102 can be replaced by a scalar function @xmath103 and the operator @xmath100 is then redefined as @xmath104 . our aim is to adapt the extension of hopf ( see @xcite ) to our model .
we recall that for any parameter @xmath105 there exists a scalar function @xmath106 such that @xmath107 & \hspace{-0.5cm}\bullet\quad\varphi_\mu(\boldsymbol{x})=0~\text{if $ d_\gamma(\boldsymbol{x})\ge 2\exp{(-1/\mu)}$\ , , where $ d_\gamma(\boldsymbol x):=\inf\limits_{\boldsymbol y\in\gamma}|\boldsymbol{x}-\boldsymbol{y}|$}\\[-0.5ex ] & \hspace{0.5cm}\text{denotes the distance of $ \boldsymbol x$ to $ \gamma$}\,,\\[2ex ] & \hspace{-0.5cm}\bullet\quad \text{if~~$d_\gamma(\boldsymbol x ) < 2\exp{(-1/\mu)}$\ , , $ j=1,\ldots , n\,.$ } \end{split}\;\right\}\tag{\text{ex}}\ ] ] for the construction of @xmath108 see also ( * ? ? ?
* lemma 2.4 , 2 , chapter iv ) .
+ let us define @xmath109 in the following lemma we establish bounds which are crucial for proving existence of velocity .
[ s3lem3 ] the function @xmath110 satisfies the following conditions @xmath111 and for any @xmath112 there exists sufficiently small @xmath105 such that @xmath113 the relations in are obvious .
we follow @xcite in order to show .
since @xmath114 sobolev s embedding theorem implies @xmath115 , so we get according to the properties of @xmath108 in the following bound @xmath116 defining @xmath117 we obtain from cauchy - schwarz and triangle inequalities @xmath118 @xmath119 and consequently @xmath120 applying hardy inequality ( see @xcite ) @xmath121 and using sobolev embedding @xmath89 , estimate ( [ s3eq13 ] ) becomes @xmath122 where @xmath123 from , , poincar inequality and from the fact that @xmath124 we conclude that for any @xmath112 we can choose sufficiently small @xmath105 such that @xmath125 holds .
therefore the proof of estimate is completed .
now , we take a look at the trilinear convective term @xmath126 the first term of above difference becomes small due to ( * ? ? ?
* lemma 2.3 , 2 , chapter iv ) , and it satisfies @xmath127 as long as @xmath105 is chosen sufficiently small . using hlder inequality , sobolev embedding @xmath128 yields @xmath129 which together with implies for sufficiently small @xmath105 the bound @xmath130 from and follows the desired estimate . while the general framework for linear and non - symmetric saddle point problems can be found in @xcite ,
our problem requires more attention due to its nonlinear character . setting @xmath131 ,
the weak formulation is equivalent to the following problem + find @xmath132 such that @xmath133 let us define the nonlinear mapping @xmath134 with @xmath135:=&a(\boldsymbol w+\boldsymbol g_\mu,\boldsymbol v)+c(\boldsymbol w+\boldsymbol g_\mu,\boldsymbol v)-(\boldsymbol f,\boldsymbol v)\\ & \;+n(\boldsymbol w+\boldsymbol g_\mu,\boldsymbol w+\boldsymbol g_\mu,\boldsymbol v)+d(\boldsymbol w+\boldsymbol g_\mu;\boldsymbol w+\boldsymbol g_\mu,\boldsymbol v)\ , , \end{split}\ ] ] whereby @xmath136 $ ] defines the inner product in @xmath137 via @xmath138:=(\nabla u,\nabla v)$ ] .
then , the variational problem ( [ s3eq17 ] ) reads in the space of @xmath21-weighted divergence free functions @xmath137 as follows + find @xmath139 such that @xmath140=0\quad\forall\;\boldsymbol v\in\boldsymbol{w}.\ ] ] we start our study of the nonlinear operator problem with the following lemma . [ s3lem4 ] the mapping @xmath42 defined in ( [ s3eq18 ] ) is continuous and there exists @xmath141 such that @xmath142 > 0\quad\forall\;\boldsymbol u\in\boldsymbol{w}\quad\textrm{with}\quad |\boldsymbol u|_1=r.\ ] ] let @xmath143 be a sequence in @xmath137 with @xmath144 .
then , applying cauchy
schwarz inequality and , we obtain for any @xmath145 @xmath146\right|\le \frac{1}{re}\left|\bigl(\varepsilon\nabla(\boldsymbol u^k-\boldsymbol u),\nabla\boldsymbol v\bigr)\right| + \frac{1}{re}\left|\bigl(\alpha(\boldsymbol u^k-\boldsymbol u),\boldsymbol v\bigr)\right|\\ & \quad + \left|\bigl(\beta|\boldsymbol u^k+\boldsymbol g_\mu|(\boldsymbol u^k-\boldsymbol u),\boldsymbol v\bigr)\right|+\left|\bigl(\beta ( |\boldsymbol u^k+\boldsymbol g_\mu|-|\boldsymbol u+\boldsymbol g_\mu|)(\boldsymbol u+\boldsymbol g_\mu),\boldsymbol v\bigr)\right|\\ & \quad+\left|n(\boldsymbol u^k,\boldsymbol u^k , \boldsymbol v)-n(\boldsymbol u,\boldsymbol u,\boldsymbol v)\right| + \left|n(\boldsymbol u^k-\boldsymbol u,\boldsymbol g_\mu,\boldsymbol v)\right|+\left|n(\boldsymbol g_\mu , \boldsymbol u^k-\boldsymbol u,,\boldsymbol v)\right|\\ & \le\frac{\varepsilon_1}{re}|\boldsymbol u^k-\boldsymbol u|_1|\boldsymbol v|_1 + \frac{1}{re}\|\alpha\|_{0,\infty}\|\boldsymbol u^k-\boldsymbol u\|_0\|\boldsymbol v\|_0\\ & \quad+\|\beta\|_{0,\infty}\|\boldsymbol u^k+\boldsymbol g_\mu\|_{0,4}\|\boldsymbol u^k-\boldsymbol u\|_0\|\boldsymbol v\|_{0,4 } + \|\beta\|_{0,\infty}\|\boldsymbol u+\boldsymbol g_\mu\|_{0,4}\|\boldsymbol u^k-\boldsymbol u\|_0\|\boldsymbol v\|_{0,4}\\ & \quad+\left|n(\boldsymbol u^k,\boldsymbol u^k , \boldsymbol v)-n(\boldsymbol u,\boldsymbol u,\boldsymbol v)\right| + c\|\boldsymbol u^k-\boldsymbol u\|_1\|\boldsymbol g_\mu\|_1\|\boldsymbol v\|_1\ , .
\end{split}\ ] ] the boundedness of @xmath147 in @xmath137 , , the poincar inequality , and the above inequality imply that @xmath148\right|\to 0\quad\text{as}\quad k\to\infty\qquad\forall\ , \boldsymbol{v}\in\boldsymbol{w}\,.\ ] ] thus , employing @xmath149}{|\boldsymbol v|_1}\,,\ ] ] we state that @xmath42 is continuous .
now , we note that for any @xmath150 we have @xmath151 = \frac{1}{re}\bigl(\varepsilon\nabla(\boldsymbol u+\boldsymbol g_\mu),\nabla\boldsymbol
u\bigr)+\frac{1}{re}\bigl(\alpha(\boldsymbol u+\boldsymbol g_\mu),\boldsymbol u\bigr)\\ & \quad + \bigl(\beta|\boldsymbol u+\boldsymbol g_\mu|(\boldsymbol u+\boldsymbol g_\mu),\boldsymbol u\bigr ) + n(\boldsymbol u+\boldsymbol g_\mu,\boldsymbol u+\boldsymbol g_\mu,\boldsymbol u)-(\boldsymbol f,\boldsymbol u)\\ & \ge\frac{\varepsilon_0}{re}|\boldsymbol u|_1 ^ 2-\frac{\varepsilon_1}{re}|(\nabla\boldsymbol g_\mu,\nabla\boldsymbol u)| + \frac{1}{re}(\alpha\boldsymbol u,\boldsymbol u)-\frac{1}{re}|(\alpha\boldsymbol g_\mu,\boldsymbol u)|\\ & \quad + ( \beta|\boldsymbol u+\boldsymbol g_\mu|,|\boldsymbol u|^2)-\left|(\beta |\boldsymbol u+\boldsymbol g_\mu|\boldsymbol g_\mu,\boldsymbol u)\right|\\ & \quad + n(\boldsymbol u,\boldsymbol g_\mu,\boldsymbol u)+n(\boldsymbol g_\mu,\boldsymbol g_\mu,\boldsymbol u)-\|\boldsymbol f\|_0\|\boldsymbol u\|_0\\ & \ge\frac{\varepsilon_0}{re}|\boldsymbol u|_1 ^ 2-\frac{\varepsilon_1}{re}|\boldsymbol g_\mu|_1|\boldsymbol u|_1\\ & \quad -\frac{1}{re}\|\alpha\|_{0,\infty}\|\boldsymbol g_\mu\|_0\|\boldsymbol u\|_0 -\left|(\beta |\boldsymbol u+\boldsymbol g_\mu|\boldsymbol g_\mu,\boldsymbol u)\right|\\ & \quad -\left|n(\boldsymbol u,\boldsymbol g_\mu,\boldsymbol u)\right| -c\|\boldsymbol g_\mu\|_1 ^ 2\|\boldsymbol u\|_1 -\|\boldsymbol f\|_0\|\boldsymbol u\|_0\ , .
\end{split}\ ] ] from the poincar inequality , we infer the estimate @xmath152 which together with , and results in @xmath153\ge\left\{\frac{\varepsilon_0}{re}-\delta(1+\|\beta\|_{0,\infty})\right\}|\boldsymbol u|_1 ^ 2\\ & \quad -\bigl\{\frac{\varepsilon_1}{re}|\boldsymbol g_\mu|_1+c_1\frac{1}{re}\|\alpha\|_{0,\infty}\|\boldsymbol g_\mu\|_0
+ \delta \|\beta\|_{0,\infty}\|\boldsymbol g_\mu\|_0 + c_2\|\boldsymbol g_\mu\|_1 ^ 2+c_3\|\boldsymbol f\|_0\bigr\}|\boldsymbol u|_1 . \end{split}\ ] ] choosing @xmath154 such that @xmath155 and @xmath156 with @xmath157 leads to the desired assertion ( [ s3eq20 ] ) .
the following lemma plays a key role in the existence proof .
[ s3lem5 ] let @xmath158 be finite - dimensional hilbert space with inner product @xmath136 $ ] inducing a norm @xmath159 , and @xmath160 be a continuous mapping such that @xmath161>0\quad\textrm{for}\quad\|x\|=r_0>0.\ ] ] then there exists @xmath162 , with @xmath163 , such that @xmath164 see @xcite .
now we are able to prove the main result concerning existence of velocity .
[ s3thm6 ] the problem ( [ s3eq19 ] ) has at least one solution @xmath150 .
we construct the approximate sequence of galerkin solutions .
since the space @xmath137 is separable , there exists a sequence of linearly independent elements @xmath165 .
let @xmath166 be the finite dimensional subspace of @xmath137 with @xmath167 and endowed with the scalar product of @xmath137 .
let @xmath168 , be a galerkin solution of ( [ s3eq19 ] ) defined by @xmath169=0,\quad\forall\ ; j=1,\ldots , m\,.\end{aligned}\ ] ] from lemma [ s3lem4 ] and lemma [ s3lem5 ] we conclude that @xmath170=0\quad\forall\;\boldsymbol w\in\boldsymbol x_m\ ] ] has a solution @xmath171 .
the unknown coefficients @xmath172 can be obtained from the algebraic system ( [ s3eq23 ] ) . on the other hand , multiplying ( [ s3eq23 ] ) by @xmath172 , and adding the equations for @xmath173 we have @xmath174\\ & \ge \left\{\frac{1}{re}-\delta(1+\|\beta\|_{0,\infty})\right\}|\boldsymbol u^m|_1 ^ 2\\ & \quad -\bigl\{\frac{1}{re}|\boldsymbol g_\mu|_1+c_1\frac{1}{re}\|\alpha\|_{0,\infty}\|\boldsymbol g_\mu\|_0
+ \delta \|\beta\|_{0,\infty}\|\boldsymbol g_\mu\|_0 + c_2\|\boldsymbol g_\mu\|_1 ^ 2+c_3\|\boldsymbol f\|_0\bigr\}|\boldsymbol u^m|_1 . \end{split}\ ] ] this gives together with ( [ s3eq22 ] ) the uniform boundedness in @xmath137 @xmath175 therefore there exists @xmath150 and a subsequence @xmath176 ( we write for the convenience @xmath177 instead of @xmath178 ) such that @xmath179 furthermore , the compactness of embedding @xmath89 implies @xmath180 taking the limit in ( [ s3eq24 ] ) with @xmath181 we get @xmath182=0\quad\forall\;\boldsymbol w\in\boldsymbol x_m.\ ] ] finally , we apply the continuity argument and state that ( [ s3eq25 ] ) is preserved for any @xmath87 , therefore @xmath12 is the solution of ( [ s3eq19 ] ) .
for the reconstruction of the pressure we need inf - sup - theorem [ s3thm7 ] assume that the bilinear form @xmath183 satisfies the inf - sup condition @xmath184 then , for each solution @xmath12 of the nonlinear problem ( [ s3eq19 ] ) there exists a unique pressure @xmath185 such that the pair @xmath186 is a solution of the homogeneous problem ( [ s3eq17 ] ) .
see ( * ? ? ?
* theorem 1.4 , 1 , chapter iv ) .
we end up this subsection by proving the existence of the pressure .
[ s3thm8 ] let @xmath187 be solution of problem .
then , there exists unique pressure @xmath185 .
we verify the inf - sup condition ( [ s3eq26 ] ) of theorem [ s3thm7 ] by employing the isomorphism of theorem [ s3thm1 ] . from (
* corollary 2.4 , 2 , chapter i ) follows that for any @xmath188 in @xmath189 there exists @xmath190 in @xmath191 such that @xmath192 with a positive constant @xmath193 . setting @xmath194 and applying the isomorphism in theorem [ s3thm1 ]
, we obtain the estimate @xmath195 where @xmath196 . from the above estimate we conclude the inf - sup condition .
we exploit a priori estimates in order to prove uniqueness of weak velocity and pressure .
[ s3thm9 ] if @xmath197 , @xmath198 are sufficiently small , then the solution of ( [ s3eq19 ] ) is unique .
assume that @xmath199 and @xmath200 are two different solutions of ( [ s3eq17 ] ) . from in lemma [ s3lem2 ]
we obtain @xmath201 .
then , we obtain @xmath202\\ & = a(\boldsymbol u_1-\boldsymbol u_2,\boldsymbol u_1-\boldsymbol u_2)+c(\boldsymbol u_1-\boldsymbol u_2,\boldsymbol u_1-\boldsymbol u_2)-(\boldsymbol f,\boldsymbol u_1-\boldsymbol u_2)\\ & \quad + n(\boldsymbol u_1+\boldsymbol g_\mu,\boldsymbol u_1+\boldsymbol g_\mu,\boldsymbol u_1-\boldsymbol u_2)-n(\boldsymbol u_2+\boldsymbol g_\mu,\boldsymbol u_2+\boldsymbol g_\mu,\boldsymbol u_1-\boldsymbol u_2)\\ & \quad + ( \beta |\boldsymbol u_1+\boldsymbol g_\mu|(\boldsymbol u_1+\boldsymbol g_\mu),\boldsymbol u_1-\boldsymbol u_2)\\ & \quad-\bigl(\beta |\boldsymbol u_2+\boldsymbol g_\mu|(\boldsymbol u_2+\boldsymbol g_\mu),\boldsymbol u_1-\boldsymbol u_2)\\ & \ge \frac{\varepsilon_0}{re}|\boldsymbol u_1-\boldsymbol u_2|_1 ^ 2-\|\boldsymbol f\|_{-1}\|\boldsymbol u_1-\boldsymbol u_2\|_1\\ & \quad + n(\boldsymbol u_1-\boldsymbol u_2,\boldsymbol u_2+\boldsymbol g_\mu,\boldsymbol u_1-\boldsymbol u_2)\\ & \quad + \bigl(\beta|\boldsymbol u_1+\boldsymbol g_\mu|(\boldsymbol u_1-\boldsymbol u_2),\boldsymbol u_1-\boldsymbol u_2\bigr)\\ & \quad + \bigl(\beta(|\boldsymbol u_1+\boldsymbol g_\mu|-|\boldsymbol u_2+\boldsymbol g_\mu|)(\boldsymbol{u}_2+\boldsymbol{g}_\mu),\boldsymbol u_1-\boldsymbol u_2\bigr)\\ & \ge \frac{\varepsilon_0}{re}|\boldsymbol u_1-\boldsymbol u_2|_1 ^
2-\|\boldsymbol f\|_{-1}\|\boldsymbol u_1-\boldsymbol u_2\|_1\\ & \quad -\left|n(\boldsymbol u_1-\boldsymbol u_2,\boldsymbol u_2,\boldsymbol u_1-\boldsymbol u_2)\right| -\left|n(\boldsymbol u_1-\boldsymbol u_2,\boldsymbol g_\mu,\boldsymbol u_1-\boldsymbol u_2)\right|\\ & \quad -\|\beta\|_{0,\infty}\left|\bigl(|\boldsymbol u_1+\boldsymbol g_\mu|\cdot |\boldsymbol u_1-\boldsymbol u_2|,|\boldsymbol u_1-\boldsymbol u_2|\bigr)\right|\\ & \quad -\|\beta\|_{0,\infty}\left|\bigl(\bigl||\boldsymbol u_1+\boldsymbol g_\mu|-|\boldsymbol u_2+\boldsymbol g_\mu|\bigr|\cdot |\boldsymbol{u}_2+\boldsymbol{g}_\mu|,|\boldsymbol u_1-\boldsymbol u_2|\bigr)\right|\ , .
\end{split}\ ] ] from cauchy - schwarz inequality and sobolev embedding @xmath203 we deduce @xmath204 @xmath205 and according to ( [ s3eq7 ] ) we have @xmath206 ) we can find @xmath207 such that @xmath208 and @xmath209 .
testing the equation ( [ s3eq17 ] ) with @xmath12 results in @xmath210 from sobolev embedding @xmath89 we deduce for sufficiently small @xmath197 @xmath211 putting ( [ s3eq28])-([s3eq32 ] ) into ( [ s3eq27 ] ) and using the inequality @xmath212 we obtain @xmath213 for sufficiently small @xmath197 , @xmath214 the constant @xmath215 in gets small and consequently the right hand side of ( [ s3eq33 ] ) is nonnegative .
this implies @xmath216 and according to theorem [ s3thm8 ] is @xmath217 .
in this section , we provide an example of the flow problem in packed bed reactors with numerical solutions at small and relatively large reynolds numbers to show the nonlinear behavior of the velocity solutions .
let the reactor channel be represented by @xmath218 where @xmath219 and @xmath220 . ] in all computations we use the porosity distribution which is determined experimentally and takes into account the effect of wall channelling in packed bed reactors @xmath221 where @xmath222 .
the distribution of the porosity is presented in figure [ fig_poro ] .
we distinguish between the inlet , outlet and membrane parts of domain boundary @xmath19 , and denote them by @xmath223 , @xmath224 and @xmath225 , respectively .
let @xmath226 at the inlet @xmath223 and at the membrane wall we prescribe dirichlet boundary conditions , namely the plug flow conditions @xmath227 and @xmath228 whereby @xmath229 , @xmath230 . at the outlet @xmath224 we set the following outflow boundary condition @xmath231 where @xmath232 denotes the outer normal . in order to avoid discontinuity between the inflow and wall conditions
we replace constant profile by trapezoidal one with zero value at the corners .
our computations are carried out on the cartesian mesh using biquadratic conforming and discontinuous piecewise linear finite elements for the approximation of the velocity and pressure , respectively .
the finite element analysis of the brinkman - forchheimer - extended darcy equation will be conducted in the forthcoming work .
the plots of velocity magnitude in fixed bed reactor ( @xmath233 ) are presented along the vertical axis @xmath234 . in the investigated reactor
the inlet velocity is assumed to be normalized ( @xmath235 ) . due to the variation of porosity
we might expect higher velocity at the reactor walls @xmath225 .
this tunnelling effect can be well observed in figure [ fig4 ] which shows the velocity profiles for different reynolds numbers .
we remark that the maximum of velocity magnitude decreases with increasing reynolds numbers .
in this work , we have extended the existence and uniqueness of solution result in literature for the porous medium flow problem based on the nonlinear brinkman - forchheimer - extended darcy law .
the existing result is valid only for constant porosity and without the considered convection effects , and our result holds for variable porosity and it includes convective effects .
we also provided a numerical solution to demonstrate the nonlinear velocity solutions at moderately large reynolds numbers for which case the brinkman - forchheimer - extended darcy law applies .
j. l. lions .
, gauthier - villars , paris , 1969 .
t. zhao . ,
ph.d thesis , university of oxford , 2014 .
, ph.d thesis , imperial college london , 2015 .
a. grillo , m. carfagna , and s federico .
, theoret .
teopm7 , vol.41 , no.4 , 283 - 322 , belgrade 2014 .
w. sobieski , a. trykozko . , technical sciences 17(4 ) , 321335 , 2014 .
a. s. lal and a. c menon .
, ijtel , issn : 2319 - 2135 , vol.3 , no.4 , 545 - 548 , 2014 . | the nonlinear brinkman - forchheimer - extended darcy equation is used to model some porous medium flow in chemical reactors of packed bed type .
the results concerning the existence and uniqueness of a weak solution are presented for nonlinear convective flows in medium with nonconstant porosity and for small data .
furthermore , the finite element approximations to the flow profiles in the fixed bed reactor are presented for several reynolds numbers at the non - darcy s range . *
2010 mathematics subject classification ( msc ) : * 76d03 , 35q35 * keywords : * brinkman - forchheimer equation , packed bed reactors , existence and uniqueness of solution + |
intense and superintense laser beam interaction with dense matter is characterized by one prominent phenomenon , that is the generation of superthermal high energy electrons and ions .
it leads to the spontaneous question as to the effects that generate them in the absorption process of intense monochromatic light beams .
latest when the kinetic temperature reaches @xmath7 ev in the plasma , @xmath8 ion charge , collisional absorption is ineffective and other effects of non - collisional nature have to become active in order to ensure absorption .
the best known non - collisional candidate so far was resonance absorption at oblique laser incidence @xcite .
it consists in the direct conversion of laser light into an electron plasma wave resonantly excited at the critical electron density where the laser frequency @xmath9 equals the plasma frequency @xmath10 .
high intensity laser pulses in the intensity domain @xmath11 with good contrast ratio are so fast rising that there is no time to form a preplasma in front of an irradiated solid sample that could couple to a resonantly excited plasma wave . therefore the search begun for new collisionless absorption processes .
the first successful proposal was the so - called @xmath12 heating @xcite .
the authors could show by particle - in - cell ( pic ) simulations that at normal incidence the lorentz force induces non - resonant electron oscillations at @xmath13 normal to the target surface which lead to appreciable absorption , target heating and production of superthermal electrons at any density above critical .
however , no attempt was made to explain how the observed absorption , i.e. , irreversibility , comes into play .
not long after a step forward was made by brunel @xcite in understanding high - power collisionless absorption . he could show after introducing a few modifications that the resonance absorption concept could be adapted to steep highly overdense plasma profiles and significant absorption could be achieved under oblique incidence despite total absence of plasma resonance at @xmath14 ( `` not - so - resonant , resonant absorption '' @xcite ) and no possibility for a plasma wave to propagate into a shallow preplasma in front of the target . instead ,
now the energy imparted to the electrons is transported into the target and deposited there .
the assumptions of cold ( i ) infinitely dense plasma ( ii ) with discontinuous interface to vacuum ( iii ) brunel could formulate the laser - matter interaction dynamics in the vacuum in front of the target in terms of three ordinary differential equations . perhaps for this reason brunel s mechanism of the electrons pulled out into the vacuum and then pushed back into the field - free target interior is frequently identified with the term `` vacuum heating '' , an expression coined later for the particles forming a kind of thermal cloud at the vacuum - ion interface in pic simulations @xcite . to be more specific ,
the reason why the term heating played an ominous role in the past and partially still does presently is because , often invoked as the leading collisionless absorption mechanism , it has never been defined properly . to introduce some rating in this respect
it seems that two groups of authors can be distinguished . by the concept of vacuum heating the first group addresses the electrons in front of the sharp - edged target that circulate in the vacuum and do not cross the interface during one laser cycle @xcite .
identification of vacuum heating with brunel s mechanism is made by the second group @xcite to contrast with anomalous skin layer absorption @xcite . by the latter all motion
is strictly confined to the target inside .
sometimes vacuum heating is interpreted as a consequence of unspecified wave breaking @xcite .
meanwhile it has been clarified that brunel / vacuum heating prevails distinctly on skin layer absorption and that vacuum heating in the restricted sense , i.e. , the contribution to absorption of the electrons not entering the target with the periodicity of the laser , is almost insignificant @xcite . from brunel s nonrelativistic model@xcite ; ( b ) @xmath15 .
energy @xmath16 in units of mean oscillation energy in vacuum ; energy cut off is at @xmath17.,scaledwidth=30.0% ] only recently a detailed analysis and discussion of brunels s model has been given @xcite .
with a view on the means of the present paper a compact summary of the results may be of interest .
the laser field component perpendicular to the target is assumed to have the structure @xmath18 .
it generates electron jets of periodicity @xmath19 .
all of them are ejected during the first quarter period and all , except @xmath20 , return to the target during the second half period @xmath21 . during the remaining @xmath22 period no further electron ejection is possible as a consequence of partial screening by the outer layers and driver field inversion .
contrary to a common believe that all electrons in one jet are pushed back by the inverted field , only half of them , lifted in the interval @xmath23 , are in phase with the driver , the other half experience a weakened driver due to screening and fall back to the target , attracted by the immobile ions , before the laser field has changed direction .
this leads quite naturally to a classification into energetic and less energetic electrons .
if not specified differently , throughout the paper we define , somehow arbitrarily , those electrons as hot whose return energy exceeds the quiver energy @xmath24 of the free electron .
accordingly , 34 % of the brunel electrons are hot and carry 82 % of the energy in the single jet .
in contrast to pic simulations the brunel spectrum is non - maxwellian with a pronounced maximum at @xmath17 followed by a sharp cut off ( see fig.[brun ] ) .
absorption @xmath25 is considerably lower than measured at intermediate of incidence @xcite but reaches unity at @xmath26 of incidence .
@xmath27 and the absorbed energy scale like @xmath28 and @xmath29 , respectively .
evidently this is the price brunel pays for oversimplification .
we want to stress that in brunel s model crossing of layers among each other during the laser action is excluded , except a few front layers whose contribution to absorption is negligible . in other words ,
the electron flow dynamics is laminar , no wave breaking or , more appropriate in the context , no breaking of flow occurs .
brunel s model offered , within limits , the first physical explanation of @xmath12 heating at @xmath13 . and yet , brunel s model does not give a direct physical feeling for the absorption process .
the numerous wrong , at least inexact interpretations of it that are still around are a direct indicator ( example : `` all electrons are pushed back by the laser field '' contrasts with the true `` free fall '' of half of them ) .
+ from an early one - dimensional ( 1d ) pic simulation the energy scaling @xmath30 has been extracted for the hot electrons @xcite ; it has been re-confirmed by independent simulations @xcite and apparently by experiments @xcite
. however , the scaling seems to be questionable for its too strong dependence on intensity ; it contrasts with other experiments @xcite and more sophisticated analysis @xcite . in turn
, corrections to the latter have been given recently on the basis of a relativistic kinematic model @xcite .
analogous scaling laws have been proposed by numerous other authors @xcite .
nevertheless there is no convergence towards a definite scaling@xcite . in order to achieve further progress
the investigation has to start from a discussion of collisionless absorption , a flow analysis of the absorbed energy into the various plasma components , a definition of the hot electron component , and completed by analytical modeling in combination with concomitant simulations .
in what follows we present our considerations on the degree of understanding collisionless laser beam absorption , the process of fast electron generation and their interaction with the low electron energy component in order to arrive at more firmly validated scaling relations in forthcoming work .
the analysis will enable us also to get insight into shortcomings of the existing models for intense laser - dense matter interaction .
let us consider the phenomenon of collisionless absorption of high - power laser beams from a more fundamental point of view .
quasi - steady state conditions poynting s theorem averaged over one laser cycle reduces to @xmath31 the energy flux density is the poynting vector @xmath32 ; it relates to the laser intensity by @xmath33 .
with @xmath34 the electron density and @xmath35 the mean electron flow velocity the current density is @xmath36 .
equation ( [ poynting ] ) describes all kinds of absorption , collisional , noncollisional , classical or quantized ; in the latter case the current density and the electric field @xmath37 are to be substituted by their operators acting on the corresponding state vector @xmath38 . in the intense laser field , despite the high particle densities involved , the classical picture is an excellent approximation .
if the laser field evolves in time as @xmath39 the current density follows as @xmath40 and @xmath41 dephasing between driver field and current determines the degree of absorption .
thus , collisionless absorption reduces to the problem of finding out which effects lead to a finite phase shift in @xmath42 . in collisional absorption it is the friction originating from the collisions between electrons and ions , @xmath43 @xmath44 collision frequency . at @xmath45 the collisional phase shift vanishes and any finite @xmath46
can only be of dynamic origin .
up to @xmath47 this dynamic origin is found in the space charge induced by @xmath48 .
the space charge generates an electrostatic field that , superposed to the laser field , determines the electron motion and leads to the desired finite phase shift @xmath46 .
this can be seen most immediately with a constant electric field @xmath49 .
it yields per electron @xmath50 it is interesting to note and it can be formally shown , however it is also physically evident that @xmath51 ; all work is done by the driver field , the space charge field @xmath52 is inert , it provides for the phase shift only .
some authors may attribute absorption to the brunel like abrupt reduction of the laser wave amplitude in the skin layer .
due to this asymmetry the energy gained by an electron in the vacuum can not be given back anymore to the wave when entering the evanescent region .
however , for this picture to work an electrostatic field component is needed , too ; transverse and longitudinal components can not be isolated from each other . in the standard resonance absorption at the critical density it is the space charge field of the electron plasma wave that provides for collisionless absorption up to @xmath53 through a phase shift @xmath54 @xcite . on the fundamental level of eqs .
( [ poyntaver ] ) and ( [ phaseshift ] ) collisionless absorption of superintense ultrashort laser pulses may be classified as fully understood now for @xmath55 for optical wavelengths .
+ all kinds of difficulties and complications arise when the degree of absorption has to be quantified
. this step can only be done by introducing appropriate models .
numerous attempts into this direction have been undertaken with the intention to explain ( i ) the origin of the hot and the warm electron components , ( ii ) when and how they are created , during one laser period by direct resonant and nonresonant acceleration , or by stochastic processes over several laser cycles , and ( iii ) where is absorption localized , in vacuum or in the skin layer .
correspondingly , the existing absorption models may be characterized as statistic or as dynamic .
examples of the first class are vacuum heating in the restricted sense @xcite , wave breaking@xcite , skin layer absorption , e.g.@xcite , linear and nonlinear landau damping@xcite .
candidates of the second class are , first of all , sharp edge absorption@xcite , longitudinal@xcite and transverse@xcite ponderomotive heating , `` zero vector potential mechanism''@xcite , relativistic kinematic model@xcite , anharmonic resonance@xcite .
let our pic simulations decide on questions ( i ) - ( iii ) and to what degree statistics is involved in the dynamics induced by the laser in dense targets .
+ now , after three decades of intense studies on superintense laser - matter interaction one would expect that such basic questions ( i ) - ( iii ) as labeled above should have found a final answer .
the numerous models presented on performed experiments tell the opposite and show that no convergence has been reached yet .
it is instructive to have a look at over 100 experimental and theoretical results on the absorption degree collected up to 2009 in @xcite , ( fig . 1 in @xcite ) , in the irradiance regime from @xmath56 , to @xmath57 .
the absorption degrees range from 5% to 95% and yet at the constant irradiance of @xmath58 absorption between @xmath59 and @xmath60 is reported .
it drastically reflects the difficulties encountered in performing unambiguous experiments with all essential parameters well defined .
it is this situation that justifies still basic 1d simulations in order to learn more about which are the essential parameters defining the underlying physical processes .
for example , a good portion of difficulties and uncertainties arising in the context of hot electron scaling have their origin in different understanding of when an electron is `` hot '' .
we consider always linear polarized ( in @xmath61 direction ) laser pulses impinging under @xmath62 angle of incidence onto strongly overdense fully ionized cold hydrogen targets with initial electron density @xmath63 such that @xmath64 @xmath65m@xmath66 the interaction of the laser beam with the target is studied by 1d relativistic pic simulations using the boost technique @xcite . if not specified differently , throughout the text we call all electrons with energies @xmath16 exceeding their mean oscillation energy @xmath24 `` hot '' or `` fast '' , @xmath67 ; \hspace { 0.4 cm } a = \frac{e{\bf |\hat{a}|}}{m_ec } \nonumber\ ] ] @xmath68 @xmath69 maximum vector potential amplitude
. the energy spectra @xmath70 of the hot electrons assume the typical shape of fig.[gauss ] , here for an @xmath71 laser pulse of full width 50 cycles and @xmath72 after 30 , 35 , 40 , and 50 laser cycles .
the straight lines of @xmath73 starting from the energy @xmath16 between @xmath74 for @xmath75 and from @xmath76 for @xmath77 are a clear signature of a maxwellian type distribution function , @xmath78 , @xmath79 boltzmann constant ( for example , the genuine nonrelativistic maxwellian contains the degeneracy factor @xmath80 to be subtracted from @xmath73 to yield a straight line , see related argument in sec .
v ) . from the slope
an electron `` temperature '' @xmath81 is determined , with the significance of @xmath82 the mean energy if the straight lines are extrapolated down to @xmath83 . in the figure @xmath81
is indicated in units of mev . for @xmath75 one
notices already cooling by energy transfer to the cold electrons and beginning plasma expansion during laser irradiation .
similar hot electron spectra have been reported by other authors , for example by @xcite . at a )
30 , b ) 35 , c ) 40 and d ) 45 cycles after the beginning of the interaction .
a @xmath84 laser pulse of peak amplitude @xmath72 and full width of 50 cycles impinges under @xmath62 onto a hydrogen target with electron density @xmath63 such that @xmath85@xmath2m@xmath66 the pulses are identical in all four frames .
target thickness varies from 40 to 60 @xmath86 vertical dashed lines mark the mean oscillation energies . the hot electrons follow a maxwellian distribution .
the maximum mean energies @xmath82 for @xmath87 are by the factors 1.04 , 1.09 , 1.7 higher than the associated @xmath24 .
@xmath82 increases during the evolution of laser pulse for @xmath88 , for @xmath75 it decreases .
power scaling @xmath89 not detected . ]
the interaction of intense laser beams with dense targets is very complex and rich of peculiar facets . on the other hand ,
consequences of basic effects , like collisionless interaction under non - harmonic resonance , are not clarified as they should . here
we report on phenomena which we believe will survive in 2d and 3d also .
let us tentatively identify vacuum heating with the energy gained by all brunel - like electrons ( `` brunel electrons '' ) .
it is the sum of energies gained during the excursion into vacuum .
this energy fraction is identified as `` vacuum heating '' and compared with the energy absorbed by all electrons .
the target thickness is chosen such that no particles are reflected from the target backside and falsify the statistics .
the laser beam intensity rises during one laser cycle to its full intensity , is subsequently held constant for 30 cycles and then sinks to zero during another full cycle . in fig.[jets ] the energies of all brunel particles and of all particles that have crossed the skin layer at a depth of half a vacuum wavelength are plotted at their crossing time for @xmath90 and @xmath91 .
the salient feature is their jet like structure predicted by the brunel model . at the low intensity ( @xmath90 )
there is a clear distinction between the brunel electrons and all electrons having undergone heating .
the increase of the energy maxima from @xmath92 to @xmath93 and the more diffuse energy profiles of the jets at half wavelength in depth is a clear indication that some heating is localized in the skin layer , in contrast to the brunel mechanism . at high intensity ( @xmath91 )
the jets assume a pronounced double structure due to the increased @xmath94 heating operating at @xmath13 , but the patterns of the two groups appear equally diffuse .
the increase in energy of the fastest electrons is almost no longer visible ( increase by @xmath95 ) .
for @xmath96 the fraction of brunel electrons results always higher than the fraction of electrons moving inward and crossing the boundary at @xmath97 in the skin layer .
the reason for the difference is to seek in the accumulation of brunel electrons in the skin layer with increasing laser cycle number , randomized there and repeatedly crossing the target - vacuum interface before disappearing in the depth of the target .
as function of time ( units in laser cycles ) for @xmath90 and @xmath91 .
double structure is due to @xmath94 acceleration
. strong reduction of @xmath98 with increasing @xmath99 is noticeable.,title="fig:",scaledwidth=24.0% ] as function of time ( units in laser cycles ) for @xmath90 and @xmath91 .
double structure is due to @xmath94 acceleration .
strong reduction of @xmath98 with increasing @xmath99 is noticeable.,title="fig:",scaledwidth=24.0% ] it is instructive to analyze the spectral distribution function @xmath70 of the brunel electrons and all electrons just when crossing positions @xmath100 and @xmath101 for the laser intensities corresponding to @xmath102 and 60 , see fig.[spectra ] .
surprising enough , at low intensity ( @xmath90 ) and , to a minor degree , also at @xmath103 the brunel electrons from the pic simulations resemble much brunel s analytical spectrum from fig.[brun ] : the sharp cut offs and the adjacent maxima of @xmath70 are reproduced , their positions however lie at much lower energies .
the maximum of @xmath70 is still clearly visible for @xmath104 ( not in the figure ) , this time at @xmath76 , but the sharp cut off changes into a transition extending over @xmath105 .
the formation of a maxwellian tail in the fast electron spectrum occurs in the skin layer and even deeper inside the target . from @xmath106 on no difference in the spectra from brunel and total electrons can be observed , they are all `` thermalized '' . at @xmath107 the spectra
extend up to @xmath108 , at @xmath91 the maximum energy is shifted to @xmath109 .
this is in agreement with the dependence of the fast electron number on laser intensity , see following section * iv.a*. from fig.[jets ] we conclude that at moderate intensities ( @xmath110 ) the skin layer contributes sensitively to the production of the most energetic electrons , either by laser - space charge resonance and/or by stochastic brunel electron - plasmon interactions . of brunel electrons and of electrons crossing positions @xmath111 and @xmath112 at intensities
@xmath113 and @xmath114 the distributions are taken after 37 laser cycles when all electrons have returned to position @xmath115 .
the non - maxwellian structure of the brunel electrons is preserved up to @xmath104 . for higher intensities
there is almost no difference between electrons heated in the vacuum and additional heating in the skin layer region.,title="fig:",scaledwidth=23.0% ] of brunel electrons and of electrons crossing positions @xmath111 and @xmath112 at intensities @xmath113 and @xmath114 the distributions are taken after 37 laser cycles when all electrons have returned to position @xmath115 .
the non - maxwellian structure of the brunel electrons is preserved up to @xmath104 . for higher intensities
there is almost no difference between electrons heated in the vacuum and additional heating in the skin layer region.,title="fig:",scaledwidth=23.0% ] of brunel electrons and of electrons crossing positions @xmath111 and @xmath112 at intensities @xmath113 and @xmath114 the distributions are taken after 37 laser cycles when all electrons have returned to position @xmath115 .
the non - maxwellian structure of the brunel electrons is preserved up to @xmath104 . for higher intensities
there is almost no difference between electrons heated in the vacuum and additional heating in the skin layer region.,title="fig:",scaledwidth=23.0% ] of brunel electrons and of electrons crossing positions @xmath111 and @xmath112 at intensities @xmath113 and @xmath114 the distributions are taken after 37 laser cycles when all electrons have returned to position @xmath115 .
the non - maxwellian structure of the brunel electrons is preserved up to @xmath104 . for higher intensities
there is almost no difference between electrons heated in the vacuum and additional heating in the skin layer region.,title="fig:",scaledwidth=23.0% ] plasma density and velocity distributions as functions of time are the natural outcome in standard pic simulations .
additional insight in the heating mechanism is obtained from the orbits @xmath116 of randomly chosen electrons .
we have analyzed numerous such computer runs each with 200 trajectories stochastically selected from ( i ) all particles heated by the laser and ( ii ) from the set of the hot electrons only . in fig.[orbits ] their time histories are depicted for the intensities @xmath103 ( left ) and @xmath91 ( right ) .
the salient features characteristic of the two groups are the following : + ( 1 ) heating of the energetic electrons is well localized at the vacuum - target interface and takes place during one laser cycle or a fraction of it .
this excludes stochastic heating of the hot electrons .
only a low fraction of them gets the high energy in the skin layer without ever emerging into vacuum .
+ ( 2 ) contrary to the standard assumption the `` slow '' return current is highly irregular as a consequence of the interaction of the jets from fig.[jets ] with the background .
it is clearly recognized in fig.[orbits ] that irregular flow sets in just with the arrival of the first jets and it becomes the more irregular the more jets it is exposed to .
the jets are accompanied by strong localized electrostatic fields that force electrons from the return current to reverse their direction towards the back of the target or , if they succeed to cross the charge cloud of an incoming jet they are heavily accelerated towards the target front to interact further with the laser field . in short words , the laminar flow of the return current
is heavily perturbed by the cherenkov emission of the plasmons excited by the jets .
the stochastic interaction , both , return electron acceleration and deceleration , has been observed in test particle models in the past @xcite .
+ ( 3 ) the plasma flow in the skin layer breaks ( like `` wave breaking '' ) , i.e. , the orbits cross each other , in contrast to brunel s laminar model of infinite target density .
+ ( 4 ) excursion into vacuum ( `` vacuum heating '' ) of the energetic electrons decreases continuously with @xmath99 increasing to reach a minimum at around @xmath107 and then to increase again . owing to the significance of effects ( 1 ) - ( 3 ) for localization and understanding hot electron generation , and understanding collisionless laser beam absorption in general
, we analyze further the acceleration process . , for @xmath103 ( a ) and @xmath91 ( c )
. the lower figures ( b ) and ( d ) show the same number of stochastically chosen orbits from the hot electrons only with @xmath117.their analysis shows acceleration , i.e. heating almost during one laser cycle or a fraction of it in the laser field at the target front and their strong interaction with the return current.,title="fig:",scaledwidth=25.0% ] , for @xmath103 ( a ) and @xmath91 ( c ) .
the lower figures ( b ) and ( d ) show the same number of stochastically chosen orbits from the hot electrons only with @xmath117.their analysis shows acceleration , i.e. heating almost during one laser cycle or a fraction of it in the laser field at the target front and their strong interaction with the return current.,title="fig:",scaledwidth=25.0% ] in brunel s model the density of the target is assumed infinite .
consequently all electrons start from the same position and no crossing of orbits occurs , the particle flow into vacuum and back to the target is laminar . despite brunel s oversimplification
his model explains basic properties of the collisionless interaction : formation of steady state jets ( fig.[jets ] ) , two groups of electron energies ( energetic electrons co - moving with the laser field , slow free fall electrons ) , majority of fast electrons stemming from the excursion into vacuum , dominant fraction of laser energy delivered to hot electrons . if therefore `` vacuum heating '' is identified with brunel s mechanism it acquires a precise meaning .
however there is the missing link to the physics of acceleration in this simple model . not to forget that in brunel s model all heated electrons are lifted into vacuum only during the first quarter laser period
the reality with skin layer included is different : the phase for lifting is stochastic , as expected from broken flow ; period doubling , tripling , quadrupling , etc .
, of electron oscillations occurs in the skin layer ( see fig.[orbits ] ) ; acceleration to high energies is a resonance effect . to see this
we must concentrate once more in detail on single orbits selected statistically . in fig.[ruhl ]
the time history of four particles starting from different depth in the target together with the electric field ( white traces ) they `` see '' during their motion is depicted for @xmath90 , i.e. , the orbits @xmath118 and the momenta @xmath119 normal to the target .
resonant interaction in the first 2 pictures is clearly recognized by the abrupt changes in @xmath118 and @xmath120 . out of resonance
the phase difference between field and momentum is @xmath121 , see @xmath120 and laser field ( white line ) in the first two pictures .
the transition to resonance , i.e. field and momentum antiparallel , occurs during half a cycle or less in the kink of @xmath118 , seen best by zooming fig.[ruhl ] .
the essential point of this resonance is its anharmonic character .
in contrast to the harmonic oscillator in the oscillator with anharmonic potential resonance is an attractor : given an excitation by the periodic laser above a certain threshold transition to resonance is unavoidable .
the reason for this behavior is as follows .
the harmonic potential is the only one in which the degree of excitation does not change its periodicity and therefore it either is driven in or out of resonance .
the average stochastically perturbed space charge potential of the plasma is flatter than harmonic and so , depending on the excitation level its eigenfrequency changes continuously from the high level @xmath10 at low excitation down below the laser frequency @xmath9 . at the crossing point resonance
it has two consequences : ( i ) driver , when in phase with the electron displacement transforms it into a runaway particle in general@xcite ; ( ii ) the resonant phase switch forces the electron to move against the bulk , the plasma flow breaks .
breaking of flow or wave breaking , respectively , often invoked as acceleration or absorption mechanism @xcite@xmath122@xcite is never their origin , it is their consequence .
+ we conclude that at @xmath123 the majority of energetic electrons is produced by resonant interaction of the laser field with the longitudinal space charge field over a fraction of one laser cycle in the vacuum as well as in the skin layer . however , there is also indirect acceleration of stochastic nature of the target background , evidenced by the last two pictures in fig.[ruhl ] with electrons heated stochastically by the plasmons emanating from the jets .
the rapidly oscillating stochastic field of the plasmons increases with the number of jets produced ; its influence on stochastic acceleration of electrons is evident in the last two pictures of fig.[ruhl ] .
( left ) and momentum @xmath120 ( right ) , white traces : electromagnetic / electrostatic field at the particle s position .
primary interaction is by resonance between transverse and longitudinal field during a fraction of laser cycle .
particles in the last two pictures experience stochastic acceleration by plasmons only.,scaledwidth=40.0% ]
as seen in the previous section * iii * there are several thermalizing mechanism : breaking of flow , skin layer noise , cherenkov plasmons from jets . as a consequence
one would expect that such effects dominate the low energy component of the electrons and that this should propagate mainly normally to the target .
the more energetic an electron is the more it feels the lorentz force in @xmath124 direction forcing its motion into laser beam direction , @xmath62 in this paper . for the single free electron starting from rest in a traveling plane wave the maximum energy gain @xmath125 and the lateral angular spread @xmath126 of the velocity component @xmath127 in propagation direction to the velocity component in the @xmath37 field direction @xmath128 are@xcite @xmath129 thus confirming this tendency . in fig.[direction ]
the momenta @xmath130 parallel to the target vs @xmath131 along the target normal of the heated electrons are depicted for @xmath132 .
the corresponding distributions of the momenta @xmath131 normal to the target over the space coordinate are shown in fig.[distribution ] .
vs @xmath133 for @xmath134 and 100 ( left picture @xmath135 : with radiation damping , right : without ) at the end of the standard laser pulse .
electron energies : @xmath136 within inner black circle , @xmath137 e_{\rm os}$ ] within dashed circle , @xmath138 e_{\rm
os}$ ] within bold circle , @xmath139 ( `` hot electrons '' ) outside . circles in the last pictures are very small ; therefore see the two insets ) .
the color of the particles in this pictures indicates their number according to the color bar .
low energy electrons within the inner circle ( majority in number ) penetrate the target normally , energetic electrons follow the laser beam direction ( dashed black lines ) , in @xmath140 also along the reflected laser beam.,title="fig:",scaledwidth=24.0% ] vs @xmath133 for @xmath134 and 100 ( left picture @xmath135 : with radiation damping , right : without ) at the end of the standard laser pulse .
electron energies : @xmath136 within inner black circle , @xmath137
e_{\rm os}$ ] within dashed circle ,
@xmath138 e_{\rm os}$ ] within bold circle , @xmath139 ( `` hot electrons '' ) outside .
circles in the last pictures are very small ; therefore see the two insets ) .
the color of the particles in this pictures indicates their number according to the color bar .
low energy electrons within the inner circle ( majority in number ) penetrate the target normally , energetic electrons follow the laser beam direction ( dashed black lines ) , in @xmath140 also along the reflected laser beam.,title="fig:",scaledwidth=24.0% ] vs @xmath133 for @xmath134 and 100 ( left picture @xmath135 : with radiation damping , right : without ) at the end of the standard laser pulse .
electron energies : @xmath136 within inner black circle , @xmath137 e_{\rm
os}$ ] within dashed circle ,
@xmath138 e_{\rm os}$ ] within bold circle , @xmath139 ( `` hot electrons '' ) outside .
circles in the last pictures are very small ; therefore see the two insets ) .
the color of the particles in this pictures indicates their number according to the color bar .
low energy electrons within the inner circle ( majority in number ) penetrate the target normally , energetic electrons follow the laser beam direction ( dashed black lines ) , in @xmath140 also along the reflected laser beam.,title="fig:",scaledwidth=24.0% ] vs @xmath133 for @xmath134 and 100 ( left picture @xmath135 : with radiation damping , right : without ) at the end of the standard laser pulse .
electron energies : @xmath136 within inner black circle , @xmath137
e_{\rm os}$ ] within dashed circle ,
@xmath138 e_{\rm os}$ ] within bold circle , @xmath139 ( `` hot electrons '' ) outside .
circles in the last pictures are very small ; therefore see the two insets ) .
the color of the particles in this pictures indicates their number according to the color bar .
low energy electrons within the inner circle ( majority in number ) penetrate the target normally , energetic electrons follow the laser beam direction ( dashed black lines ) , in @xmath140 also along the reflected laser beam.,title="fig:",scaledwidth=24.0% ] vs @xmath133 for @xmath134 and 100 ( left picture @xmath135 : with radiation damping , right : without ) at the end of the standard laser pulse .
electron energies : @xmath136 within inner black circle , @xmath137
e_{\rm os}$ ] within dashed circle ,
@xmath138 e_{\rm os}$ ] within bold circle , @xmath139 ( `` hot electrons '' ) outside .
circles in the last pictures are very small ; therefore see the two insets ) .
the color of the particles in this pictures indicates their number according to the color bar .
low energy electrons within the inner circle ( majority in number ) penetrate the target normally , energetic electrons follow the laser beam direction ( dashed black lines ) , in @xmath140 also along the reflected laser beam.,title="fig:",scaledwidth=25.0% ] vs @xmath133 for @xmath134 and 100 ( left picture @xmath135 : with radiation damping , right : without ) at the end of the standard laser pulse .
electron energies : @xmath136 within inner black circle , @xmath137
e_{\rm os}$ ] within dashed circle , @xmath138
e_{\rm os}$ ] within bold circle , @xmath139 ( `` hot electrons '' ) outside .
circles in the last pictures are very small ; therefore see the two insets ) .
the color of the particles in this pictures indicates their number according to the color bar .
low energy electrons within the inner circle ( majority in number ) penetrate the target normally , energetic electrons follow the laser beam direction ( dashed black lines ) , in @xmath140 also along the reflected laser beam.,title="fig:",scaledwidth=25.0% ] from the pictures it is not directly seen that the majority of slow electrons move in the direction of the target normal ; however , as expected from ( [ velocities ] ) , with increasing energy the electrons follow indeed the direction of the laser beam .
in addition , at @xmath91 and 100 an appreciable percentage is accelerated into specular direction .
the reduction of the absolute number of hot electrons with increasing intensity , their almost vanishing at @xmath141 and their impressive reappearance towards @xmath135 is particularly striking . this effect will have direct impact on every attempt to formulate scaling laws for the `` hot electron '' production .
we have counted their number as a function of intensity ; the result is reported in table i. drop and increase with intensity is beyond expectation .
.number of hot electrons per unit area ( arbitrary units ) in dependence of @xmath99 for @xmath142 [ cols= " < , < , < , < , < , < , < " , ] [ table : second ] the increase in ion energy beyond @xmath104 is due to the deeper penetration of the laser as a consequence of the recession of the electrons by the radiation pressure and hence increased energy coupling to the ions . the overall absorption drops continuously with increasing laser intensity .
in contrast to the runaway energy @xmath143 in ( [ velocities ] ) the free quiver energy at fixed oscillation center is @xmath144 \sim i^{1/2}$ ] .
however this reduction is counterbalanced by the relativistic increase of the critical density , @xmath145 . as the speed of the moderately hot electrons approaches @xmath146 the absorption into energetic electrons , which is the major portion , should not change ; the drop must have a different , nonrelativistic origin .
although the scaling of @xmath24 and @xmath147 may be oversimplified ( see @xcite for @xmath147 scaling ) it is correct in its tendency .
our current explanation attributes the very pronounced reduction of absorption to the limiting effect of the electrostatic field on the oscillation amplitude of the single electron : with increasing intensity @xmath148 the electrons are pushed more and more inward by the radiation pressure .
the electron oscillating in the neighborhood of the vacuum - ion interface oscillates in a narrow anharmonic potential the half width of which towards the target interior is a small fraction of the wavelength ( `` profile steepening '' ) . a similar reduction of absorption has been reported for normal incidence , with an explanation that agrees qualitatively with ours @xcite .
latest beyond @xmath149 absorption by the fastest electrons increases again .
they are runaway electrons .
the electrostatic potential is strong but finite .
the phase at which the electrons enter the laser beam is stochastic . within them
there will some of them happen to be in resonance with the field and subject to the doppler shift @xmath150 with @xmath151 lorentz factor , @xmath152 wave vector .
if such an electron is moving inward from the vacuum it sees the incident wave at a doppler downshifted low frequency and is accelerated over a longer distance whereas the reflected wave is seen at a highly upshifted frequency and represents merely a high frequency disturbance . for an electron moving outward towards the vacuum the accelerating field
is that of the reflected wave .
the proof of this acceleration mechanism is based on the study of single particle motions , and is directly confirmed by the appearance of energetic electrons flowing into vacuum in the reflected wave direction in the picture for @xmath91 and @xmath135 of fig.[direction ] . to give a numerical example of electron displacement lengthening @xmath153 in a plane tn
: sa laser wave ( @xmath154 nm ) during a forth cycle @xmath155 : @xmath156 at @xmath157 and @xmath158 at @xmath159 . for comparison , at @xmath160 this shift is 0.03 only
. beyond @xmath161 radiation reaction on the electron motion has to be taken into account @xcite .
a summary of absorption into all plasma channels ( electrons , ions , plasmons ) its fraction into electrons , the decrease of absorption towards a pronounced minimum close to zero at @xmath162 and its rise beyond is presented in fig.[absorption ] .
from intensity scaling the experimentalist and theoretician expect analytical formulas of the shape of the electron spectrum as a function of the laser intensity .
as such a goal seems to be beyond reach at present the high power laser community has limited its focus on the energetic electrons . there , the generation of a maxwellian tail is one of the characteristics of high power interaction .
it is also the most interesting part of the spectrum because , as seen from table ii it contains the main part of the absorbed energy and , last but not least , it is relevant to applications for collective ion acceleration , radiation sources , medical applications , and others .
it is aimed at how the number of energetic electrons , the degree of absorption and the mean energy scale with intensity . on the basis of present knowledge scaling of the first two quantities
is not feasible . regarding the mean energy , or the hot temperature @xmath163 , respectively , despite the frequent attempts in experiment and theory
no convergence has been achieved so far at all . in the light of our foregoing analysis
there is not much surprise about . + .
the reduction is due to oscillation inhibition by the induced electrostatic field , its rise beyond is mainly a consequence of entrainment ( `` runaway electrons '' ) .
+ , scaledwidth=40.0% ] the frequently invoked ponderomotive scaling ( `` wilks scaling'')@xcite of @xmath164 is based on the idea that each laser cycle energetic electrons with energy average in the range of about @xmath165 are generated .
subsequently this scaling has been recognized as too strong and , in first place guided by experiments @xcite , has been replaced by the milder power law @xcite @xmath166 in the intensity range @xmath167 .
it is intended as to be applied to the unidirectional maxwellian electrons reaching the analyzer at time @xmath168 .
the search for the right mean energy scaling is equivalent to the search for the process of absorption . in the absence of the stochastic element inherent in collisions it is important to understand whether a maxwellian tail is one of the signatures of the interaction and ,
if it is , why .
to arrive at a maxwellian distribution in an ensemble it is sufficient to know that , given a certain amount of particles @xmath169 containing the amount of energy @xmath170 , all possible states in the relevant phase space are equally likely and that the hamiltonian is given by the sum @xmath171 which expresses the property that the single electrons are uncorrelated .
if the relevant phase space is @xmath172 , as for example in statistical thermodynamics , the resulting distribution is the maxwellian momentum distribution @xmath173 , in disagreement with fig.[gauss ] .
however , as outlined in the foregoing chapters , fast electron generation is by anharmonic resonance .
this has the important properties : ( i ) resonance is an attractor for all electrons above a certain oscillation energy , in contrast to harmonic resonance ; the always present crossing of trajectories is a clear indicator of it .
( ii ) all @xmath169 electrons have the same chance to resonate anharmonically regardless of their phase with respect to the laser driver .
this makes it very likely that the relevant phase space is the energy acquired at resonance rather than the momentum .
then from ( i ) and ( ii ) follows that collisionless absorption is accompanied by a maxwellian tail of energetic electrons ( consequence of anharmonic attractor ) and the spectrum scales like @xmath174 without a degeneracy factor @xmath175 this is what we also deduce from fig.[gauss ] . with laser intensity @xmath176 , standard pulse .
stars : @xmath177 ( `` hot electrons '' ) ; diamonds : @xmath178 ( `` warm electrons '' ) .
in contrast to fig.[gauss ] here @xmath179 is the average taken over the single energies @xmath180 .
solid line : scaling after @xcite , @xmath181 $ ] . dashed line : scaling after @xcite , @xmath182^{0.34}$ ] .
+ , scaledwidth=40.0% ] let us first examine fig.[gauss ] for pulses @xmath71 in the intensity range @xmath183 .
the uncertainties on the mean energy ( slope of log scale ) in pictures a ) ( 30 cycles ) and c ) ( 40 cycles ) are considerable , nevertheless we can conclude with certainty that neither wilks original ponderomotive scaling@xcite ( @xmath184 ) nor its improved version are met to some extent .
they are far too weak .
however , the analysis shows that the assumption @xmath185 with the constant @xmath186 not far from unity works .
this means that at these relatively low intensities ( from @xmath187 to @xmath75 ) the scaling is + @xmath188 in agreement with @xcite . from @xmath189 on @xmath24
is well approximated by @xmath190 . in fig.[scal2 ] we extended the search for scaling from @xmath75 up to @xmath91 the latter is already in the runaway regime of absorption .
satisfactory agreement with @xcite is obtained for @xmath148 from @xmath191 to @xmath57 . beyond the change in the absorption mechanism and the stiffer coupling to the ions
is noticeable in the increase of slope relative to @xcite .
the focus of the present paper is on the physics of collisionless absorption of intense laser beams in dense targets in the intensity domain @xmath192 for optical wavelengths , on the variation of the spectral composition of the energetic electrons with intensity and on their scaling with the latter .
most remarkable results are the brunel like spectral hot electron distribution at the relativistic threshold , the minimum of absorption at @xmath193 , the drastic reduction of the number of hot electrons in this domain and their reappearance beyond , the strong coupling with the return current beyond expectation , and a strong hot electron scaling in @xmath194 , a scaling in vague accordance with current published estimates in @xmath195 and a strong increase beyond . on a fundamental level understanding collisionless absorption
is equivalent to the search for the non - orthogonality of induced current density to the laser field .
the answer is found in the interplay of the laser field with the space charge field induced by it .
the idealized model of brunel works already on this basis .
it is capable of explaining important effects at the relativistic intensity threshold and below , like the generation of two groups of electrons , a hot and a cold component . the non - maxwellian spectrum predicted by the model is found in our simulations at the relativistic threshold and below ; with increasing @xmath99 it is washed out .
by following test orbits we are able to localize absorption at the vacuum - target interface and skin layer for all intensities below the radiation reaction limit at @xmath196 in linear polarization , in agreement with brunel for non relativistic intensities .
if therefore the ominous `` vacuum heating '' is invoked as responsible for absorption this is correct if it is identified with brunel s mechanism .
what it does not explain , and brunel does not either , is the underlying physics , i.e. the phase shift and , in concomitance , orbits crossing .
an explanation in terms of physics has to show that ( i ) such a breaking of flow is not by accidence and ( ii ) a hot maxwellian tail in the spectrum is a natural outcome from strong drivers .
anharmonic resonance is currently the best model explaining both aspects .
it rules stochastic heating and `` weave breaking '' out automatically .
anharmonic resonance constitutes an attractor ( fix point ) .
this kind of resonance always happens in presence of a sufficiently strong driver at any laser frequency and any target density , in contrast to harmonic resonance which is bound to @xmath197 .
when crossing resonance the momentum of an electron undergoes a phase shift by @xmath198 or a fraction of it with respect to the bulk of the plasma .
wave breaking , here more appropriately called breaking of flow owing to profile steepening on lengths of a small fraction of a laser wavelength , is a consequence of absorption and energetic electron generation , not its origin . from @xmath199 anharmonic resonance
is strengthened by the generation of runaway electrons due to trapping in both , the incident and the reflected laser wave .
the reduction up to nearly disappearance of hot electrons @xmath117 is attributed to oscillation inhibition by the ponderomotive space charge field . in the whole intensity domain considered the major fraction of laser energy is deposited in the hot and moderately hot electrons .
the next significant portion goes into cherenkov plasmons excited by the periodic plasma jets . from the analysis of the test trajectories their coupling to the neutralizing return current is apparent .
we consider it as an important aspect when modeling anomalous transport of heat and fast electrons in compressed matter .
finally , we reexamined the hot electron scaling in 1d , perhaps the most controversially discussed subject in the pertinent literature .
acceptable coincidence with the leading approximations is only found in the intensity range @xmath200 .
the deviations below @xmath201 are to be attributed mainly to the imprecise proportionality between @xmath24 and @xmath202 . in the runaway absorption regime
the governing scaling law is still to be discovered .
the main reason for the current misunderstandings and disagreements have to be attributed to the poor knowledge of the electron energy spectrum @xmath70 .
what is missing most at present in the experiment and in the theory is a clear definition of what means `` hot '' and `` cold '' electrons . in order not to fall into this deficiency
we define electrons as `` hot '' and `` moderately hot '' if their energy is higher than @xmath24 and @xmath203 .
it is approximately the range of the maxwellian tail .
the commonly used terminology `` maxwellian '' is misleading because it refers to the electron velocities @xmath35 or momenta with a distribution law @xmath204 .
this differs from our findings ( and , implicitly , others ) of a boltzmann distribution @xmath205 for the relevant restricted phase space of total energies @xmath206 .
this work was in part supported by the dfg within the sfb 652 .
pic simulations were performed using the computing resources granted by the john von neumann - institut for computing ( research center jlich ) under the project hro01 .
we gratefully thank dr .
andrea macchi ( university of pisa , italy ) for providing us his 1d pic code and prof .
su - ming weng ( jiao tong university , shanghai ) for his contributions of ideas and suggestions . the continuous support of this work by prof . dieter bauer ( university of rostock , germany ) is gratefully acknowledged .
9 a. d. piliya , sov .
* 11 * , 609 ( 1966 ) .
kull , phys .
fluids * 26 * , 1881 ( 1983 ) . w. l. kruer , k. estabrook , phys .
fluids * 28 * , 430 ( 1985 ) .
f. brunel , phys .
lett . * 59 * , 52 ( 1987 ) ; phys . fluids * 31*,2714 ( 1988 ) .
p. gibbon , a. r. bell , phys .
* 68 * , 1535 ( 1992 ) .
p. gibbon , phys .
lett . * 73 * , 664 ( 1994 ) .
l. m. chen , j. zhang , q. l. dong , h. teng , t. j. liang , l. z. zhao , and z. y. whei , phys .
plasmas * 8 * , 2925 ( 2001 ) ; q. dong and j.zhang , science in china * 46 * , 71 ( 2003 ) ; d. umstadter , j. phys .
d * 36 * , r151 ( 2003 ) ; s. kato , j. plasma fusion res . * 6 * , 658 ( 2004 ) .
a. v. getz and v. p. krainov , j. exp .
* 101 * , 80 ( 2005 ) .
w. rozmus and v. t. tikhonchuk , phys.rev .
a * 42 * , 7401 ( 1990 ) ; w. rozmus , v. t. tikhonchuk , and r. cauble , phys .
plasmas * 3 * , 360 ( 1996 ) .
b. yang , w. l. kruer , r. m. more , and a. b. langdon , phys .
plasmas * 2 * , 3146 ( 1995 ) ; t .- y .
b. yang , w. l. kruer , a. b. langdon , and t. w. johnston , phys .
plasmas * 3 * , 2702 ( 1996 ) .
g. ferrante , m. zarcone , and s. a. uryupin , phys .
plasmas * 9 * , 4560 ( 2002 ) .
s. kato , b. bhattacharyya , a. nishiguchi and k. mima , phys .
fluids b * 5 * , 564 ( 1993 ) .
cai , phys .
plasmas * 13 * , 063108 ( 2006 ) .
d. bauer and p. mulser , phys .
plasmas * 14 * , 023301 ( 2007 ) .
p. mulser , s
weng , and t.liseykina , phys .
plasmas * 19 * , 043301 ( 2012 ) .
m. cerchez , r. jung , j. osterholz , t. toncian , o. willi , p. mulser , and h. ruhl , phys .
lett . * 100 * , 245001 ( 2008 ) .
s. c. wilks , w. l. kruer , m. tabak , and a. b. langdon , phys .
lett . * 69 * , 1383 ( 1992 ) .
t. baeva , s.gordienko , a.p.l.robinson , p. a.norreys , phys .
plasmas * 18 * , 056702 ( 2011 )
. g. malka and j. l. miquel , phys .
. lett . * 77 * , 7578 ( 1996 ) .
p. mckenna , f. lindau , o. lundh , d. c. carroll , r. j. clarke , k. w. d. ledingham , t. mccanny , d. neely , a. p. l. robinson , l. robson , p. t. simpson , c .- g .
wahlstrm , and m. zepf , plasma phys .
fusion * 49 * , b223 ( 2007 ) .
f. n. beg , a. r. bell , a. e. dangor , c. n. danson , a. p. fews , m. e. glinsky , b. a. hammel , p. lee , p. a. norreys , and m. tatarakis , phys .
plasmas * 4 * , 447 ( 1997 ) .
m. g. haines , m. s. wei , f. n. beg , and r. b. stephens , phys . rev .
* 102 * , 045008 ( 2009 ) .
p. gibbon , a. a. andreev , and k. yu . platonov , plasma phys .
fusion * 54 * , 045001 ( 2012 ) c. d. chen , j. a. king , m. h. key , k. u. akli , f. n. beg , h. chen , r. r. freeman , a. link , a. j. mackinnon , a. g. macphee , p. k. patel , m. porkolab , r. b. stephens , and l. d. van woerkom , rev .
. instrum . * 79 * , 10e305 ( 2008 ) .
a. g. macphee , k. u. akli , f. n. beg , c. d. chen , h. chen , r. clarke , d. s. hey , r. r. freeman , a. j. kemp , m. h. key , j. a. king , s. le pape , a. link , t. y. ma , h. nakamura , d. t. offermann , v. m. ovchinnikov , p. k. patel , t. w. phillips , r. b. stephens , r. town , y. y. tsui , m. s. wei , l. d. van woerkom , and a. j. mackinnon , rev .
instrum . * 79 * , 10f302 ( 2008 ) .
t. kluge , t. cowan , a. debus , u. schramm , k. zeil , and m. bussmann , phys .
lett . * 107 * , 205003 ( 2011 ) .
a. macchi , _ a superintense laser - plasma interaction theory primer _ ( springer new york , 2013 ) , chap .
y. sentoku , v.y .
bychenkov , k. flippo , a. maksimchuk , k. mima , g. mourou , z.m .
sheng , and d. umstadter , appl .
b * 74 * , 207 ( 2002 )
. v. s. rastunkov , v. p. krainov , laser phys . * 15 * , 262 ( 2005 ) .
d. f. zaretsky , ph .
a. korneev , s. v. popruzhenko , and w. becker , j. phys .
b * 37 * , 4817 ( 2004 ) . ph . a. korneev , s. v. popruzhenko d. f. zaretsky , and w. becker , laser phys . lett . * 2 * , 452 ( 2005 ) . a. v. sofronov and v. p. krainov , j. phys .
b * 37 * , l329 ( 2004 ) .
a. pukhov , rep .
phys . * 66 * , 47 ( 2003 ) .
p. mulser , d. bauer , and h. ruhl , phys .
* 101 * , 225002 ( 2008 ) .
j. r. davies , plasma phys.control .
fusion * 51 * , 014006 ( 2009 ) .
methods phys .
a * 544 * , 61 ( 2005 ) .
p. gibbon , a. a. andreev , e. lefebvre , g. bonnaud , h. ruhl , j. delettrez , and a.r .
bell , phys . of plasmas
* 6 * , 947 ( 1999 ) .
a. j. kemp , y. sentoku , and m. tabak , phys .
e * 79 * , 066406 ( 2009 ) .
p. mulser and d. bauer , _ high power laser - matter interaction _
( springer heidelberg , 2010 ) , p. 356 . m. sherlock , e.g. hill , r.g .
evans , and s.j
. rose , phys .
113 * , 255001 ( 2014 ) .
s. m. weng , p. mulser , and z. m. sheng , phys .
plasmas * 19 * , 022705 ( 2012 ) .
j. sanz , a. debayle , and k. mima , phys .
* 85 * , 046411 ( 2012 ) .
h. chen , s. c. wilks , w. kruer , p. patel , and r. shepherd , phys .
plasmas * 16 * , 020705 ( 2009 ) .
n. naumova , t. schlegel , v. t. tikhonchuk , c. labaune , i. v. sokolov , and g. mourou , phys .
lett . * 102 * , 025002 ( 2009 ) .
m. tamburini , t.v .
liseykina , f. pegoraro , and a. macchi , phys.rev .
e * 85 * , 016407 ( 2012 ) ; m. tamburini , f. pegoraro , a. di piazza , c. h. keitel , and a. macchi , new j. phys .
* 12 * , 123005 ( 2010 ) .
l. l. ji , a. pukhov , i. yu .
kostyukov , b. f. shen , and k. akli , phys . rev.lett * 112 * , 145003 ( 2014 ) .
s. c. wilks and w. l. kruer , ieee j. of quantum electronics , * 33 * , 1954 ( 1997 ) . | among the various attempts to understand collisionless absorption of intense and superintense ultrashort laser pulses a whole variety of models and hypotheses has been invented to describe the laser beam target interaction . in terms of basic physics collisionless absorption is understood now as the interplay of the oscillating laser field with the space charge field produced by it in the plasma . a first approach to this idea
is realized in brunel s model the essence of which consists in the formation of an oscillating charge cloud in the in front of the target , therefore frequently addressed by the vague term `` vacuum heating '' .
the investigation of statistical ensembles of orbits shows that the absorption process is localized at the ion - vacuum interface and in the skin layer : single electrons enter into resonance with the laser field thereby undergoing a phase shift which causes orbit crossing and braking of brunel s laminar flow . this anharmonic resonance acts like an attractor for the electrons and leads to the formation of a maxwellian tail in the electron energy spectrum .
most remarkable results of our investigations are the brunel like spectral hot electron distribution at the relativistic threshold , the minimum of absorption at @xmath0 in the plasma target with the electron density of @xmath1@xmath2m@xmath3 the drastic reduction of the number of hot electrons in this domain and their reappearance in the highly relativistic domain , and strong coupling , beyond expectation , of the fast electron jets with the return current through cherenkov emission of plasmons .
the hot electron energy scaling shows a strong dependence on intensity in the moderately relativistic domain @xmath4 , a scaling in vague accordance with current published estimates in the range @xmath5 , and again a distinct power increase beyond @xmath6 .
the low energy electrons penetrate normally to the target surface , the energetic electrons propagate in laser beam direction . |
the global topology of spacetime , which is not fixed by the equations of general relativity , plays an important role in quantum field theory even in a flat spacetime .
when a spacetime is multiply connected only those modes of a field that satisfy boundary conditions determined by the topology of the spacetime are relevant in the calculation of locally measurable quantities such as the stress - energy tensor @xmath5 .
for example , in a cylindrical two dimensional spacetime , @xmath6 , the only allowed momentum is an integer multiple of @xmath7 where @xmath8 is the circumference in the closed spatial direction .
dewitt , hart , and isham @xcite thoroughly studied the effects of multiple connectedness of the spacetime manifold ( called mbiosity ) , twisting of the field , and orientability of manifold on @xmath4 for a massless scalar field in various topological spaces . in this paper
we extend their work to an untwisted massive scalar field .
we evaluate @xmath0 in four dimensional spacetime manifolds of the type @xmath9 where @xmath10 can be either @xmath11 , @xmath12 , @xmath13 , the mbius strip @xmath14 , or the klein bottle @xmath15 .
both the mbius strip and the klein bottle are examples of nonorientable manifolds .
another issue addressed in this paper is the effect of field mass on chronology protection . in previous papers @xcite
it has been shown that @xmath4 for a massless scalar field diverges on a chronology horizon in various spacetimes with closed timelike curves ( ctc s ) .
the backreaction of the metric to this diverging stress - energy through the einstein field equations may be able to prevent formation of ctc s .
however , boulware s work indicates that @xmath4 for a massive scalar field remains finite on the chronology horizon in gott space @xcite .
we have confirmed boulware s result in grant space which is holonomic to gott space @xcite and found that @xmath16 is finite on the chronology horizon provided the mass of the scalar field is above a lower limit which depends on the topological identification scale lengths of the spacetime . in sec .
ii a general procedure for using point - splitting regularization to calculate the vacuum expectation value of @xmath5 for a free massive scalar field in a flat but multiply connected spacetime is described .
we then apply this method to spacetimes without ctc s in sec .
iii and to those with ctc s in sec . iv . throughout our calculations natural units in which @xmath17 are used and the metric signature is @xmath18 .
our calculation of the vacuum expectation value of the stress - energy tensor is greatly simplified by the facts that all curvature components vanish in a flat spacetime and that we do not need to deal with renormalization of the field mass @xmath20 and curvature coupling constant @xmath21 arising from interactions .
however , we do have to worry about the topology of the spacetime manifold since it may allow two points on the manifold to be connected by multiple geodesics . in this paper the minkowski vacuum state
is assumed to be the default vacuum state in all spacetimes considered .
this assumption is defended later by an argument based on a particle detector carried by a geodesic observer .
point - splitting regularization ( or the `` method of images '' ) is used to take multiple connectedness into account .
the differences in topology of the spacetimes will appear only in the geodesic distances between image charges and in the number of geodesics connecting the points .
once geodesical distances for a particular topology are found the calculation of @xmath19 reduces to simple differentiation of the hadamard elementary function and taking the coincidence limit .
the stress - energy tensor @xmath5 is formally defined as a variation of the action with respect to the metric . in a flat four dimensional spacetime the stress - energy tensor for a general free scalar field is given by @xmath22 note that @xmath4 depends upon the value of @xmath21 even when the curvature vanishes .
for conformal coupling @xmath23 ; for minimal coupling @xmath24 .
we will allow arbitrary values of @xmath21 to make our results as general as possible .
the scalar field @xmath25 satisfies the klein - gordon equation @xmath26 .
since every term in @xmath5 is quadratic in the field variable @xmath27 , we can split the point @xmath28 into @xmath28 and @xmath29 and take the coincidence limit as @xmath30 .
@xmath31\left\{\phi(x ) , \phi(\tilde{x})\right\ } , \label{eq2}\end{aligned}\ ] ] where @xmath32 is the anticommutator of @xmath33 and @xmath34 .
covariant derivatives @xmath35 and @xmath36 are to be applied with respect to @xmath28 and @xmath29 .
we have also symmetrized @xmath5 over @xmath27 and @xmath37 . before taking the vacuum expectation value of @xmath5
we need to define the vacuum state of the spacetimes .
this is nontrivial because some spacetimes lack a global timelike killing vector field , which is required to define positive frequency .
the spacetimes we will consider may all be constructed by making topological identifications of minkowski space , with coordinates and metric given by @xmath38 in the case of the @xmath1 , @xmath2 , @xmath3 , the mbius strip and the klein bottle spacetimes , there exists a global timelike killing vector field ( i. e. , @xmath39 ) and so does a natural vacuum state . the effect of multiple connectedness on the field is only making it periodic in closed spatial directions .
also , it allows the field to be twisted , and we will briefly comment of the effect of twisting and the vacuum energy density in the next section . since each section of these spacetimes between the boundaries is identical to a portion of minkowski space , we can assume the default vacuum states of these spacetimes to be identical to that of ordinary minkowski space . as an example of a spacetime with no natural vacuum state , we will consider grant space in sec .
although grant space does have local timelike killing vector fields everywhere , they can not be patched together to form a global timelike killing vector field @xcite .
however , each section of the spacetime is identical to minkowski space , and a geodesic observer will not detect any particles if the spacetime is in the minkowski vacuum state .
hence , we can also use the minkowski vacuum state for the grant space . the vacuum expectation value of stress - energy tensor is now given by sandwiching both sides of eq .
( [ eq2 ] ) by the minkowski vacuum state @xmath40 .
@xmath41g^{(1)}(x , \tilde{x } ) , \label{eq4}\end{aligned}\ ] ] where the hadamard elementary function @xmath42 is defined as @xmath43 and satisfies the klein - gordon equation @xmath44 . in minkowski space @xmath45 is a function of the half squared geodesic distance @xmath46 between two points @xmath28 and @xmath29 and has the form @xmath47 where @xmath48 is a step function and @xmath49 and @xmath50 are modified bessel functions of the first and second kinds , respectively @xcite .
because the spacetime is multiply connected , there can be more than one geodesic connecting the two points @xmath28 and @xmath29 .
for example , suppose the spacetime is closed in the @xmath51 direction .
we can connect @xmath28 and @xmath29 with a direct path , or we can start from @xmath28 and circle around in the @xmath51 direction once , twice , or an arbitrary number of times before arriving at @xmath29 .
since the path circling around @xmath52 times can not be deformed continuously into the one which circles around @xmath53 ) times , all inequivalent paths must be taken into account .
equivalently we can consider this situation as an electrostatic problem and use the method of images .
the `` image charges '' of the point @xmath29 are located at @xmath54 , where @xmath8 is the periodicity ( or circumference ) in the closed spatial direction .
all these image charges are connected to the point @xmath28 by geodesics whose half squared distances @xmath55 are given by @xmath56 where @xmath57 is the position of the @xmath52th image charge . a contribution from each image charge
is summed over to construct the regularized hadamard function @xmath58 .
however , using @xmath58 in eq .
( [ eq4 ] ) gives infinity because the stress - energy tensor is not renormalized yet .
the infinite vacuum energy term associated with the minkowski vacuum state must be subtracted from @xmath58 .
this term comes from @xmath59 for the image charge at @xmath60 . excluding it from the summation we obtain the renormalized hadamard function @xmath61 which , using eq.([eq4 ] ) , gives the renormalized stress - energy tensor @xmath62 .
@xmath63 the calculation of @xmath4 has thus been reduced to ( 1 ) writing an appropriate @xmath55 for each topology , ( 2 ) applying the derivative operator in eq .
( [ eq4 ] ) , and ( 3 ) taking the coincidence limit as @xmath30 .
in this section the vacuum expectation value @xmath64 of the stress - energy tensor of a massive scalar field is evaluated in four dimensional spacetimes with @xmath65 topology .
a spacetime of this type has a global timelike killing vector field ( e. g. , @xmath39 ) and thus a natural vacuum state .
the first topology we investigate is @xmath66 .
the manifold can be either orientable or nonorientable . in an orientable manifold a triad obeying the right - hand rule
maintains the same handedness as it is translated around the closed spatial dimension . in cartesian
coordinates @xmath67 the spacetime is closed in the @xmath51 direction with periodicity @xmath8 , and the following points are identified : @xmath68 where @xmath52 is an integer .
the half squared geodesic distance @xmath55 between the point @xmath28 and the @xmath52th image charge at @xmath57 is equal to @xmath69 .
\label{eq10}\ ] ] we will be concerned with only a spacelike separation ( @xmath70 ) , so the first term in eq .
( [ eq6 ] ) is used for the hadamard function .
this is because the intervals between the image charges are always spacelike in all spacetimes concerned .
using the prescription for the calculation of @xmath64 described in the previous section , we obtain @xmath71\right\ } , \label{eq11}\ ] ] where @xmath72 .
@xmath4 does not depend on the curvature coupling .
the energy density @xmath73 is negative due to a fact that only certain wavelengths are allowed in the @xmath51 direction compared to minkowski space in which all wavelengths are allowed . in the massless
limit the above result reduces to that given in ref.@xcite .
figure [ fig1 ] shows a plot of @xmath74 vs. @xmath20 for @xmath75 . the effect of mass on the energy density is to raise it ( decreasing its magnitude ) because for a larger field mass , its characteristic wavelength or compton wavelength @xmath76 becomes shorter and is less sensitive to the global structure of the manifold .
the next manifold that we will consider is @xmath77 .
the spacetime is now closed in two spatial dimensions , @xmath51 and @xmath78 , with periodicities @xmath8 and @xmath79 , respectively .
the following points are identified @xmath80 where @xmath52 and @xmath81 are integers .
the half squared geodesic distance @xmath82 between @xmath28 and the @xmath83th image charge at @xmath84 is given by @xmath85 .
\label{eq13}\ ] ] following the prescription we find @xmath86\right\ } , \label{eq14}\end{aligned}\ ] ] where @xmath87 .
the summation is over all values of @xmath52 and @xmath81 except when @xmath88 simultaneously . as it can be seen in fig .
[ fig1 ] @xmath74 for @xmath2 is lower ( greater in magnitude ) than that for @xmath89 .
the increased mbiosity of the manifold lowers the energy density @xcite .
the last example of orientable spacetime manifold is the one which is closed in all spatial directions ( @xmath90 ) .
the topology of this spacetime is such that the following points are identified : @xmath91 where @xmath92 , and @xmath93 are integers and @xmath94 , and @xmath95 are the periodicities in the @xmath96 , and @xmath97 directions , respectively .
the half squared geodesic distance @xmath98 is equal to @xmath99 , \label{eq16}\end{aligned}\ ] ] and the resulting stress - energy tensor is @xmath100\right\ } , \label{eq17}\end{aligned}\ ] ] where @xmath101 .
the summation is over all possible value of @xmath92 , and @xmath93 except when @xmath102 at the same time . as we can see in fig .
[ fig1 ] the increased mbiosity lowers the vacuum energy density @xmath74 . increasing the field mass raises it toward zero .
the procedure described previous section is readily applicable to the twisted field .
the only modification required is to introduce the factor of @xmath103 in the summation in eq .
( [ eq8 ] ) .
this will make field antiperiodic .
dewitt , hart , and isham have shown that twisting of the field raises @xmath74 above the zero - value of minkowski vacuum state for a massless scalar field @xcite .
our numerical calculations indicate that this is true even for a massive scalar field .
the manifolds we have examined so far are orientable .
next , we examine how the orientation of the manifold affects the vacuum polarization . in a nonorientable manifold a triad which defines the right - hand rule flips to the one for the left - hand rule when it is transported around the closed spatial dimension .
the simplest example of a nonorientable manifold is a mbius strip ( @xmath104 ) , which locally looks like the @xmath105 manifold @xcite .
a four dimensional mbius strip spacetime @xmath106 can be constructed by identifying the points @xmath107 the half squared geodesic distance @xmath55 is equal to @xmath108 ^ 2 + ( x^3 - \tilde{x}^3)^2\right\}. \label{eq19}\ ] ] it is obvious that @xmath55 and thus @xmath19 will depend on @xmath78 even after the coincidence limit is taken .
@xmath109\right\ } \nonumber \\ & & { } + \frac{m^4}{4\pi^2}\sum_{n = -\infty}^{\infty}\left\{\left[2(1 - 2\xi)\frac{k_2(z_{2n + 1})}{(z_{2n + 1})^2 } + 4(4\xi - 1)m^2(x^2)^2\frac{k_3(z_{2n + 1})}{(z_{2n + 1})^3}\right ] \right . \nonumber \\ & & { } \times { \rm diag}[-1 , 1 , 0 , 1 ] \nonumber \\ & & \left .
{ } + m^2(2n + 1)^2 a^2\frac{k_3(z_{2n + 1})}{(z_{2n + 1})^3}{\rm diag}[0 , -1 , 0 , 0]\right\ } , \label{eq20}\end{aligned}\ ] ] where @xmath110 and @xmath111^{\frac{1}{2}}$ ] .
the first term is identical to that for the case of @xmath1 with twice the periodicity .
the periodicity is doubled because the triad must circle around in the @xmath51 direction twice in order to be oriented in the original sense .
it is interesting that the second term in eq .
( [ eq20 ] ) is dependent on the curvature coupling constant @xmath21 ; @xmath74 increases linearly with increasing @xmath21 . for the mbius strip
the nonorientability of the manifold lowers the energy density from that of @xmath1 as we can see in fig .
[ fig2 ] . along the @xmath78 direction
@xmath74 has a reverse bell shape centered at @xmath112 .
a plot of @xmath74 vs. @xmath78 is shown in fig.[fig3 ] .
the last example of a spacetime with @xmath113 topology is the klein bottle spacetime ( @xmath114 ) .
the klein bottle @xmath115 is a natural extension of the mbius strip with an additional spatial periodic boundary condition . in cartesian coordinates , the points to be identified are @xmath116 and @xmath82 is equal to @xmath117 .
\label{eq22}\end{aligned}\ ] ] again , the resulting @xmath19 depends on both @xmath78 and @xmath21 , @xmath118\biggr\ } \nonumber \\ & & { } + \frac{m^4}{4\pi^2}\sum_{n , m = -\infty}^{\infty}\left\{\left[2(1 - 2\xi)\frac{k_2(z_{2n + 1 , m})}{(z_{2n + 1 , m})^2 } \right .
\right . \nonumber \\ & & \left.\left . { } + 4(4\xi - 1)m^2(x^2 - mb)^2\frac{k_3(z_{2n + 1 , m})}{(z_{2n + 1 , m})^3}\right]{\rm diag}[-1 , 1 , 0 , 1 ] \right .
\nonumber \\ & & \left . { } + m^2 ( 2n + 1)^2 a^2\frac{k_3(z_{2n + 1 , m})}{(z_{2n + 1 , m})^3}{\rm diag}[0 , -1 , 0 , 0]\right\ } , \label{eq23}\end{aligned}\ ] ] where @xmath119 and @xmath120^{\frac{1}{2}}$ ] .
the first summation is identical to @xmath4 for @xmath2 with spatial periodicities doubled in both @xmath51 and @xmath78 directions .
the first summation simply gives a constant negative shift in the vacuum stress - energy throughout the spacetime .
the second half of @xmath4 is dependent on both the @xmath78 coordinate and the curvature coupling @xmath21 .
@xmath74 is oscillatory in the @xmath78 direction with minima at integer multiples of the periodicity @xmath79 in that direction ( see fig .
[ fig3 ] ) . as in the mbius strip spacetime
an increase in the value of @xmath21 causes @xmath74 to increase .
however , in contrast to the mbius strip spacetime , the nonorientability of the manifold decreases the magnitude of the energy density compared to the @xmath2 spacetime as we can see in fig.[fig2 ] .
the spacetimes that we have examined so far have one thing in common ; they have a clearly defined killing time coordinate @xmath121 . in those spacetimes @xmath39 is a globally defined timelike killing vector field which can be used to define a vacuum state .
however , it is possible to arrange topological identification of flat space such that a clearly defined time coordinate for the entire spacetime manifold does not exist . as an example of a spacetime without a global timelike killing vector field
, we will consider grant space .
grant space is interesting because it contains closed timelike curves ( ctc s ) .
the chronology protection conjecture , proposed by hawking @xcite , states that nature prevents the formation of ctc s in spacetimes which have an initial chronal region . in order to prove the conjecture , the protection mechanism within the laws of physics
must be identified , if it exists .
currently the most promising candidate for the protection mechanism is the backreaction on the metric due to vacuum polarization of quantized matter fields .
quantum field fluctuations pile up on top of themselves near a chronology horizon causing the vacuum expectation value of stress - energy tensor @xmath19 to grow without bound .
this diverging @xmath0 will act back on the metric through the einstein field equations and change the spacetime geometry , possibly preventing the appearance of ctc s .
the vacuum stress - energy of a massless scalar field has been shown to diverge on the chronology horizon in misner space @xcite , gott space @xcite , wormhole spacetime @xcite , and roman space @xcite . in our previous paper @xcite
we evaluated @xmath4 for a massive scalar field with arbitrary curvature coupling in misner space and found that it diverges on the chronology horizon . on the other hand
, boulware @xcite has shown that the vacuum stress - energy of a massive scalar field is finite on the chronology horizon of gott space .
we would like to know whether that is still true in the grant space , which is holonomic to gott space yet contains misner space as a special limit .
the original misner space was developed to illustrate topological pathologies associated with taub - nut ( newman - unti - tamburino ) type spacetimes @xcite .
misner space is simply the flat kasner universe with an altered topology .
its metric in misner coordinates @xmath122 is @xmath123 that misner space is flat can be easily seen ; the above metric becomes identical to the minkowski metric via the following coordinate transformation : @xmath124 grant space is constructed by making topological identifications in the @xmath125 and @xmath126 directions : @xmath127 misner space is the special case @xmath128 . in cartesian coordinates the above identification is equivalent to @xmath129 it can be shown that grant space is actually a portion of ( holonomic to ) gott space , which consists of two infinitely long , straight cosmic strings passing by each other @xcite .
the periodicities @xmath8 and @xmath79 in grant space are related respectively to the relative speed and distance between the two cosmic strings in gott space . as @xmath79 approaches zero ( the misner space limit ) the impact parameter of the two strings also approaches zero .
grant space can be considered as a portion of the @xmath2 spacetime of the previous section with the boundaries in the @xmath51 direction moving toward each other at constant velocity .
a spacetime diagram of the maximally extended grant space is shown in fig . [ fig4 ] .
in regions i and iv radial straight lines represent @xmath130 surfaces .
hyperbolas are constant @xmath131 surfaces .
all image charges of a point are located on the same hyperbolic surface .
as a particle crosses the radial boundary , @xmath130 , it is lorentz boosted in a new inertial frame moving at a speed @xmath132 in the @xmath51 direction with respect to the original frame and is translated by @xmath133 in the @xmath78 direction .
what is extraordinary about grant space is that it contains nonchronal regions ( ii and iii ) . in those regions
the roles of @xmath131 and @xmath125 are switched .
the radial boundaries are now spacelike and the spacetime becomes periodic in the time ( @xmath131 ) direction .
this topological identification allows the formation of ctc s in those regions . the boundaries ( @xmath134 ) separating chronal regions ( i and iv ) and nonchronal regions ( ii and iii ) are chronology horizons which are a kind of cauchy horizons .
the chronological structure of grant space is discussed in ref .
@xcite . within each interval between the periodic boundaries ( i. e. , one period )
@xmath39 can play the role of a timelike killing vector field , but it is impossible to define a global timelike killing vector field by patching these @xmath39 s together . without a global timelike killing vector field the vacuum state of the spacetime can not be defined .
however , we argue that the minkowski vacuum state is a valid vacuum state of the grant space .
each interval in grant space is identical to a portion of minkowski space , so a geodesic observer in the interval will not detect any particle in the minkowski vacuum . since the only difference between one interval to its neighbors is a constant relative velocity in the @xmath51 direction and a translation in the @xmath78 direction , geodesic observers in the neighboring intervals will not find any particles in the same vacuum state .
the state in which no geodesic observer detects any particles can be considered as a vacuum state .
this allows us to use the same renormalized hadamard function eq.([eq6 ] ) in the calculation of @xmath19 in grant space .
the result of the calculation is most simply expressed in the misner coordinates , ( @xmath135 ) , @xmath136 \frac{k_2(z_n)}{z_n^2 } , \nonumber \\
\left\langle 0\left|t_1 ^ 1\right|0\right\rangle & = & \frac{m^4}{2\pi^2}\sum_{n = 1}^{\infty}\left[1 + 4\xi\sinh^2\left(\frac{na}{2}\right)\right ] \nonumber \\ & & { } \times \left[\frac{k_2(z_n)}{z_n^2 } - 4m^2 ( y^0)^2\sinh^2\left(\frac{na}{2}\right)\frac{k_3(z_n)}{z_n^3 } \right ] , \nonumber \\
\left\langle 0\left|t_2 ^ 2\right|0\right\rangle & = & \frac{m^4}{2\pi^2}\sum_{n = 1}^{\infty}\left\{\left[1 + 2(4\xi - 1)\sinh^2\left(\frac{na}{2}\right)\right]\frac{k_2(z_n)}{z_n^2 } \right .
\nonumber \\ & & \left . { } + m^2\left[4(1 - 4\xi)(y^0)^2\sinh^4\left(\frac{na}{2 } \right ) - n^2 b^2\right]\frac{k_3(z_n)}{z_n^3}\right\ } , \nonumber \\
\left\langle 0\left|t_3 ^ 3\right|0\right\rangle & = & \left\langle 0\left|t_{22}\right|0\right\rangle + \frac{m^6 b^2}{2\pi^2 } \sum_{n = 1}^{\infty}n^2\frac{k_3(z_n)}{z_n^3 } , \label{eq28}\end{aligned}\ ] ] where @xmath137^{\frac{1}{2}}$ ] .
the trace is equal to @xmath138 \frac{k_2(z_n)}{z_n^2 } \right .
\nonumber \\ & & \left .
{ } + \left[-z_n^2 + 8(1 - 6\xi)m^2(y^0)^2\sinh^4\left(\frac{na}{2}\right)\right ] \frac{k_3(z_n)}{z_n^3}\right\}. \label{eq29}\end{aligned}\ ] ] on the chronology horizon ( @xmath139 ) , the components of the stress - energy tensor are @xmath136 \frac{k_2(mnb)}{(mnb)^2 } , \nonumber \\
\left\langle 0\left|t_1 ^ 1\right|0\right\rangle & = & \left\langle 0\left|t_0 ^ 0\right|0\right\rangle , \nonumber \\
\left\langle 0\left|t_2 ^ 2\right|0\right\rangle & = & \frac{m^4}{2\pi^2}\sum_{n = 1}^{\infty}\left\{\left[1 + 2(4\xi - 1)\sinh^2\left(\frac{na}{2}\right)\right ] \frac{k_2(mnb)}{(mnb)^2 } - \frac{k_3(mnb)}{mnb}\right\ } , \nonumber \\
\left\langle 0\left|t_3 ^ 3\right|0\right\rangle & = & \frac{m^4}{2\pi^2}\sum_{n = 1}^{\infty}\left[1 + 2(4\xi - 1)\sinh^2\left(\frac{na}{2}\right)\right ] \frac{k_2(mnb)}{(mnb)^2}. \label{eq30}\end{aligned}\ ] ] figure [ fig5 ] shows how the energy density @xmath74 depends on the field mass @xmath20 for the conformal coupling ( @xmath23 ) .
for @xmath140 , @xmath74 diverges on the chronology horizon .
the contribution to @xmath74 from the @xmath52th image charge for @xmath141 is proportional to @xmath142 $ ] .
the factor of @xmath143 comes from the doppler shift as the particle is boosted in the @xmath125 direction @xmath52 times . in misner space this factor causes @xmath4 to diverge on the chronology horizon , and it might prevent the formation of ctc s .
however , in grant space the exponentially decaying factor @xmath144 , which comes from the nonvanishing geodesical distances between image charges in the @xmath126 direction @xmath79 , makes @xmath16 finite for values of @xmath145 .
this result agrees with boulware s similar calculation in gott space @xcite .
this result implies that the metric backreaction may not be large enough to significantly change the grant space geometry if the field mass is sufficiently large .
in calculating the vacuum expectation value of the stress - energy tensor for a massive scalar field with arbitrary curvature coupling in multiply connected flat spacetimes , we have shown that a nonzero field mass raises the value of the energy density ( decreasing its magnitude ) and confirmed that increased mbiosity lowers the energy density .
the introduction of nonorientability can evidently either increase the magnitude of the energy density ( mbius band ) or decrease it ( klein bottle ) .
another effect on the stress - energy that we have not examined yet is self - interaction ( e. g. , @xmath146 theory ) .
the calculation of @xmath4 for a self - interacting field is much more complicated than that for a free field since it requires renormalization of quantities such as the field mass , which will be dependent on the field @xmath25 , the curvature coupling constant @xmath21 , and the field coupling constant @xmath147 . in spacetimes with relatively simple topology , such as @xmath89 and casimir - type spacetime ,
@xmath4 has been calculated by birrell and ford , ford , ford and yoshimura , kay , and toms @xcite . in a spacetime with ctc s self - interaction
is known to cause failure of unitarity @xcite . in grant space
we discovered that the divergence of @xmath16 on the chronology horizon depends on relative size of the field mass to the ratio of the periodicities .
this result may have significant consequences for chronology protection .
it suggests that the metric backreaction from the stress - energy of a massive quantized field will likely not be large enough to significantly alter the geometry and prevent the formation of ctc s .
if quantized matter fields are to provide the chronology protection mechanism , our result would indicate that only massless fields may be capable of providing a sufficiently strong backreaction to prevent the formation of ctc s . outside the domain of quantum gravity , this would place a heavy responsibility on the electromagnetic field ( and conceivably neutrino fields , should any be massless ) as the sole protector of chronology .
electronic mail address : [email protected] .
electronic mail address : [email protected] .
b. s. dewitt , c. f. hart , and c. j. isham , physica * 96a * , 197 ( 1979 )
. w. a. hiscock and d. a. konkowski , phys .
d * 26 * , 1125 ( 1982 ) .
v. p. frolov , phys .
d * 43 * , 3878 ( 1991 ) .
kim and k. s. thorne , phys .
d * 43 * , 3929 ( 1991 ) .
g. klinkhammer , phys .
d * 46 * , 3388 ( 1992 ) .
d. g. boulware , phys .
d * 46 * , 4421 ( 1992 ) .
j. d. e. grant , phys . rev .
d * 47 * , 2388 ( 1993 ) .
m. lyutikov , phys .
d * 49 * , 4041 ( 1994 ) .
t. tanaka and w. a. hiscock , phys .
d * 49 * , 5240 ( 1994 ) .
d. laurence , phys .
d. * 50 * , 4957 ( 1994 ) .
n. d. birrell and p. c. w. davies , _ quantum fields in curved space _
( cambridge university press , cambridge , england , 1982 ) , pp . 87 . s. a. fulling , _ aspects of quantum field theory in curved spacetime _ ( cambridge university , cambridge , england , 1989 ) , pp
r. m. wald , _ general relativity _
( the university of chicago press , chicago , usa , 1984 ) , pp . 363 . s. w. hawking , phys .
d * 46 * , 603 ( 1992 ) . c. w. misner , in _ relativity theory and astrophysics i : relativity and cosmology , lectures in applied mathematics _ , edited by j. ehlers ( american mathematical society , providence , 1967 ) , pp
. s. w. hawking and g. f. r. ellis , _ the large scale structure of space - time _
( cambridge university , cambridge , england , 1973 ) , pp . 170174 .
for a recent review of these notions , see k. s. thorne , in _ general relativity and gravitation 1992 _ , proceeding of the thirteenth international conference on general relativity and gravitation , edited by r. j. gleiser , c. n. kozameh , and o. m. moreschi ( institute of physics , bristol , 1993 ) , pp . 295 .
l. h. ford , proc .
r. soc .
lond . a * 368 * , 305 ( 1979 ) .
l. h. ford and tetz yoshimura , phys .
* 70a * , 89 ( 1979 ) . b. s. kay , phys .
d * 20 * , 3052 ( 1979 ) .
n. d. birrell and l. h. ford , phys .
d * 22 * , 330 ( 1980 ) .
d. j. toms , ann .
phys . * 129 * , 334 ( 1980 ) .
d. j. toms , phys .
d * 21 * , 928 ( 1980 ) .
j. friedman , m. s. morris , i. d. novikov , f. echeverria , g. klinkhammer , k. s. thorne , and u. yurtsever , phys .
d * 42 * , 1915 ( 1990 ) .
f. echeverria , g. klinkhammer , and k. s. thorne , phys .
d * 44 * , 1077 ( 1991 ) .
i. d. novikov , phys .
d * 45 * , 1989 ( 1992 ) .
j. l. friedman , n. j. papastamatiou , and j. z. simon , phys .
d * 46 * , 4442 ( 1992 ) .
j. l. friedman , n. j. papastamatiou , and j. z. simon , phys .
d * 46 * , 4456 ( 1992 ) .
m. s. morris , k. s. thorne , and u. yurtsever , phys .
lett . * 61 * , 1446 ( 1988 ) . | the vacuum expectation value of the stress - energy tensor @xmath0 is calculated in several multiply connected flat spacetimes for a massive scalar field with arbitrary curvature coupling .
we find that a nonzero field mass always decreases the magnitude of the energy density in chronology - respecting manifolds such as @xmath1 , @xmath2 , @xmath3 , the mbius strip , and the klein bottle . in grant space , which contains nonchronal regions , whether @xmath4 diverges on a chronology horizon or not depends on the field mass . for a sufficiently large mass @xmath4 remains finite , and the metric backreaction caused by a massive quantized field may not be large enough to significantly change the grant space geometry . |
most early logical systems ( for propositional and predicate logic ) allowed no hypotheses and so had no rules for introducing or cancelling them . these could be represented by a finite set of axiom schemes and rules of inference such as modus ponens and generalisation . later natural deduction systems which did allow hypotheses had fewer axiom schemes but required introduction and elimination rules for hypotheses .
herbrand showed that classical hilbert style and natural deduction style propositional and predicate logics had the same theorems ( i.e. judgements with empty contexts ) .
pure type systems ( ptss ) , defined below , have two rules that introduce hypotheses and two that cancel them . in this paper
we answer a question of fairouz kamareddine are there hilbert style ptss ? " .
when we define hilbert style ptss ( hptss ) as ptss with empty contexts , with a finite set of extra axiom schemes , with @xmath8 representing arbitrary sorts , and some extra rules , it is obvious that there are hptss
. we will be interested in whether , for ptss , there are theorem equivalent hptss .
we will answer this question for a number of classes of ptss which include all the ptss , from the standard literature , that we have examined .
the methods we use , for proving that a hpts is equivalent to a pts , are along the lines of those of herbrand , but rather more complex . just as combinator based programming languages , requiring no free , or in fact , no variables , have proved useful in practice , perhaps an hpts , which also requires no ( free ) variables , that is theorem equivalent to a pts may be useful .
also , perhaps some metatheoretical results may be proved more easily for an hpts than for the equivalent pts .
each pure type system ( pts ) @xmath9 has a set of variables @xmath10 , a set of constants @xmath11 , a set of sorts " @xmath12 .
it has a class of pseudoterms given by @xmath13 .
if @xmath14 and @xmath15 are pseudoterms , @xmath16 is a statement , @xmath17 is a context if it is a sequence of statements ; @xmath18 is then called a judgement .
a pts has a set of axioms @xmath19 each of the form @xmath20 where @xmath21 and @xmath22 .
then it has a set @xmath23 of triples @xmath24@xmath25 , which determine under what conditions a term @xmath26 is in a sort .
most ptss are known by a specification " @xmath27 ( as usually @xmath28 ) .
the pts postulates are as follows : when there are two judgements as premises in a rule , we call the left one the _
major premise _ and the right one the _ minor premise_. later we will need the following definition : @xmath29 is an _ inhabited sort _ @xmath30 if @xmath31 for some @xmath32 .
@xmath29 is a _
normal form inhabited sort _ @xmath33 if for some term @xmath32 in normal form , @xmath34 the translation @xmath35 $ ] of bunder and dekkers [ 3 ] translates the pseudoterms and statements of ptss into terms of illative combinatory logic ( icl ) as follows : @xmath36 = x,\quad[c]=c,\quad [ xy]=[x][y]\ ] ] @xmath37 = [ a][x],\quad [ \pi x{:}x.yx ] = { \bf g } [ x][y]\quad ( x\notin fv(xy))\ ] ] where @xmath38 ( @xmath4 is the combinator equivalent to @xmath39 .
terms in icl can be represented without any free variables at all using the combinators @xmath4 and @xmath3 ( equivalent to @xmath40 ) .
@xmath41 represents roughly @xmath42 or @xmath43 .
icl , designed as a foundation for logic and mathematics , has a rule like ( abstraction ) which was derived in bunder [ 2 ] from a set of axioms . in section 6
we will see how the methods developed there lead to the ones used here .
the main difference between ptss and standard icls , other than the lack of distinction between terms and types , lies in the ( abstraction ) rule .
the direct counterpart to the icl rule would have @xmath44 , for @xmath45 .
this is the most important factor in making it difficult to have equivalent hilbert - style ptss .
we define hilbert - style ptss as follows : each _ hilbert style pure type system _ ( hpts ) has @xmath10 , @xmath46 , @xmath47 , @xmath48 , statements , contexts and judgements as for ptss , except that the contexts are always empty .
a hpts has a set of sorts @xmath47 and a set of axioms @xmath19 , as for ptss , and an additional finite set @xmath49 of axiom schemes in which sort variables " can be replaced by sorts .
most hptss are known by a specification " @xmath50 ( as usually @xmath28 ) .
a hpts has the pts ( application ) and ( conversion ) rules ( with empty contexts ) as well as : note the latter rules are derivable for all ptss , for hptss neither is , even using ( conversion ) .
if @xmath9 is a pts with specification @xmath51 , a hpts @xmath52 , with specification @xmath53 will be _ equivalent _ if @xmath54 here @xmath55 stands for provability in @xmath9 and @xmath56 in @xmath52 . if the pts is arbitrary or obvious from the context we use @xmath57 and @xmath58 .
@xmath59 will a function of @xmath23 , i.e. it will include axioms such as @xmath60\:[\pi u{:}s_1.\pi v{:}(\pi x{:}u .
s_2).s_3]$ ] if @xmath61 below are some ptss that have been studied in the literature ( particularly barendregt [ 1 ] and geuvers [ 4 ] ) . in @xmath62 , @xmath63 , in all other cases
@xmath64 consists of all the constants visible in @xmath19 and @xmath23 .
@xmath65 is used as an abbreviation for @xmath66 .
the pts used in the proof assistant coq we will call @xmath67 .
it has as axioms : @xmath68 more axioms are generated by @xmath69 in early versions @xmath70 is given by @xmath71 for all @xmath72 .
coq 8.0 replaces @xmath73 by @xmath74
. we will be able to determine whether or not there are equivalent hptss for all of the above .
we now state a number of standard lemmas for ptss .
most proofs can be found in barendregt [ 1 ] or bunder and dekkers [ 3 ] .
[ l : one ] if @xmath75 , then 1 .
@xmath76 are distinct ; 2 .
@xmath77 ; 3 .
@xmath78 for @xmath79 .
[ l : two ] if @xmath80 , @xmath81 and @xmath82 then @xmath83\vdash m[x:=n]\:b[x:=n]$ ] .
[ l : three ] if @xmath84 , where @xmath85 , then @xmath86 .
[ l : four ] let @xmath87
. then 1 .
@xmath88 ; 2 .
@xmath89 ; 3 .
@xmath90@xmath91 ; 4 .
@xmath92@xmath93 ; 5 .
@xmath94 $ ] . in each case
the derivations , of the judgements of the form @xmath95 in _ ( iii ) _ to _ ( v ) _ , are shorter than that of @xmath87 .
[ l : five ] if @xmath87 then @xmath96 @xmath97 or @xmath98 $ ] .
[ l : six ] if @xmath87 , then 1 .
@xmath99 implies @xmath100 , 2 .
and @xmath101 implies @xmath18 .
[ l : seven ] if @xmath87 , then 1 .
@xmath102 implies @xmath103 , 2 .
@xmath104implies that for @xmath105 there is an @xmath106 such that @xmath107 .
the following lemma specifies a set of ptss whose axioms are its only theorems .
the equivalent hpts is then trivially one with no extra axioms , i.e. with @xmath108 .
[ l : eight ] in a pts satisfying @xmath109\bigr)\eqno{(\$)}\ ] ] we have @xmath110 .
we show @xmath111 by induction on the derivation of @xmath112 ( [ eq : ma ] ) clearly does not come by ( start ) or ( weakening ) .
if ( [ eq : ma ] ) comes by ( application ) from @xmath113 where @xmath114 and @xmath115 $ ] , we have by the induction hypothesis @xmath116 , which is impossible . if ( [ eq : ma ] ) comes by ( abstraction ) from @xmath117 where @xmath118 and @xmath119 , then by the induction hypothesis @xmath120 , which is impossible . if ( [ eq : ma ] ) comes by ( product ) from @xmath121 where @xmath122 and @xmath123 , then , by the induction hypothesis , @xmath124 , which is impossible by ( $ )
if ( [ eq : ma ] ) comes by ( conversion ) from @xmath125 then by the induction hypothesis @xmath126 .
however then @xmath32 and @xmath127 must be in normal form and so @xmath128 and @xmath129 .
if ( [ eq : ma ] ) is an axiom , the result holds trivially .
this implies the following theorem and corollary .
[ t : nine ] a pts satisfying _ ( $ ) _ has an equivalent hpts , with @xmath130 but this is trivial in that it has only its axioms as theorems .
[ c : ten ] @xmath131 and @xmath132p each have an equivalent hpts , but @xmath133 is the only theorem of both systems .
in @xmath131 and @xmath134p there is no term @xmath32 such that @xmath135 and the only theorem is @xmath136 .
we can show , by a single ( product ) rule preceeded by two uses of an axiom and a ( start ) or ( weakening ) rule , that in the other ptss , given in section 2 , there are theorems that are not axioms .
most of these are given below .
[ l : eleven ] 1 .
in @xmath137 we have @xmath138 .
2 . in @xmath139 , @xmath140 , @xmath141 , @xmath142 , @xmath143 , @xmath144 and @xmath145 , we have @xmath146 .
3 . in @xmath147 and @xmath148 , we have @xmath149 .
4 . in @xmath150 ,
@xmath151 and @xmath152 we have @xmath153 .
we now give a condition under which , in a pts , certain sorts have an infinite number of inhabitants of the form @xmath26 that are not substitution instances of each other .
we show later that many ptss with this property can not be equivalent to hptss .
[ l : twelve ] assume that in a pts there is a finite sequence @xmath154 such that : @xmath155 then @xmath156 for an infinite number of @xmath157-distinct terms @xmath26 which are not ( @xmath158 for @xmath159 ) substitution instances of each other .
assume that we have ( @xmath160 ) for @xmath161 .
as @xmath162 we have , for some @xmath163 , in normal form @xmath164 now we show , by induction on @xmath165 that , for @xmath166 , there is a @xmath167 and an @xmath168 such that @xmath169 for each @xmath165 we have , by ( @xmath160 ) , an @xmath170 such that @xmath171 and a @xmath167 such that @xmath172 when @xmath173 we have @xmath174 above , otherwise we have @xmath175 by the induction hypothesis . by ( weakening )
we have @xmath176 and by ( product ) we have ( [ eq : two ] ) .
so ( [ eq : two ] ) holds for @xmath79 and , as we have @xmath177 , @xmath178 repeating the above , with @xmath179 for @xmath163 , we get @xmath180 and similarly @xmath181 , if @xmath182 for @xmath183 , then @xmath184 and so @xmath185 and eventually @xmath186 but @xmath163 is a proper part of @xmath187 and is in normal form , which is impossible
. hence @xmath188 are @xmath157-distinct inhabitants of @xmath29 all of the form @xmath189 , which are not substitution instances of each other .
( @xmath160 ) is satisfied for many sequences @xmath190 and many ptss .
here we list one such sequence and sort for most of the ptss given in section 2 .
[ l : thirteen ] 1 .
@xmath137 , @xmath139 and @xmath140 satisfy _ ( @xmath191)_. 2 .
@xmath147 , @xmath142 , @xmath141 , @xmath148 , @xmath143 , @xmath144 and @xmath145 satisfy _
( @xmath192)_. 3 .
@xmath150 , @xmath151 and @xmath152 satisfy _ ( @xmath193)_. a. by lemma [ l : eleven](i ) , ( ii ) with @xmath194 . b. for @xmath141 , by lemma [ l : eleven](ii ) , with @xmath195 . for the others with @xmath196 . c. by lemma [ l : eleven](iv ) with @xmath196 .
now we can prove the main result in the section .
[ t : fourteen ] if , in a pts @xmath9 , _
( @xmath197 ) _ holds for some @xmath198 and @xmath199 then there is no hpts equivalent to @xmath9 . by lemma [ l : twelve ] , if ( @xmath197 ) holds we have , for an infinite number of @xmath157 - distinct terms @xmath26 , which are not substitution instances of each other @xmath200 suppose that there is an equivalent @xmath52 .
as a hpts has only a finite set of axioms @xmath59 , at least some must be derived , in @xmath52 , by ( application ) and perhaps ( conversion ) , ( type reduction ) and ( subject reduction ) from @xmath201 where @xmath202 and @xmath203 \rightarrow_{\beta } s_1 $ ] . by the equivalence of @xmath9 and @xmath52 also : @xmath204 so by correctness of types ( lemma [ l : five ] ) , for some @xmath205 @xmath206 and by the generation lemma ( lemma [ l : four](iii ) ) we have : @xmath207 where @xmath208 . now by the substitution lemma ( lemma [ l : two ] ) , ( [ eq : three ] ) and , if needed , subject reduction ( lemma [ l : six ] ) @xmath209 . by lemma [ l : four](i )
this contradicts ( @xmath210 ) , so @xmath9 has no hpts equivalent .
[ t : fifteen ] @xmath137 , @xmath140 , @xmath211 , @xmath147 , @xmath141 , @xmath148 , @xmath143 , @xmath150 , @xmath151 , @xmath152 , @xmath144 and @xmath145 have no equivalent hptss . by lemma [ l : thirteen ] and theorem [ t : fourteen ] .
note that ( @xmath212 ) is not satisfied by any sort @xmath29 in @xmath213 and @xmath132coq . for @xmath131 ( @xmath214 ) holds but ( @xmath215 ) does not for any @xmath216 for @xmath132p ( @xmath214 ) fails , ( @xmath217 ) holds but ( @xmath218 ) does not for any @xmath216
in implicational logic the @xmath219-introduction rule is @xmath220 .
the hypothesis a in @xmath221 is cancelled in @xmath222 .
this rule is proved in a hilbert - style system by induction on the number of steps in a derivation that allows hypotheses .
we assume that an hypothesis can be cancelled in the previous step ( or steps ) and use this to show it can be cancelled in the next . in intuitionistic and classical implicational logic three cases
are needed and each requires the hilbert - style system to have a particular axiom or theorem . if the hypothesis @xmath223 is itself the step in the deduction we need @xmath224 if the deduction step is an axiom or another hypothesis than @xmath223 we need @xmath225 if the deduction step comes by modus ponens from @xmath226 we need @xmath227 note that the three theorems we require represent the simple types of the combinators @xmath2 , @xmath3 and @xmath4 ( when @xmath219 is replaced by @xmath228 ) . in illative combinatory logic ,
the introduction rule for @xmath229 ( restricted generality ) is @xmath230 , where @xmath231 is a constant , @xmath232 and @xmath233 is the hypothesis being cancelled . in the proof of this rule in a hilbert - style system,(see bunder[2 ] ) , the first two cases are similar to those for the proof of implicational introduction .
the third is the case where @xmath234 is derived from @xmath235 and @xmath236 .
again , by induction , we assume that the @xmath229-introduction step can be applied to the previous steps . the axioms of the hilbert - style system , when rewritten with @xmath237 for @xmath238 are : @xmath239\cr \vdash\ldots\ & [ ( a\rightarrow b\rightarrow a)\,{\bf k}]\cr \vdash\ldots\ & [ ( ( a\rightarrow b \rightarrow c)\rightarrow ( a\rightarrow b)\rightarrow a \rightarrow c)\,{\bf s}]\cr } \ ] ] where @xmath240 represent conditions involving @xmath231 on @xmath241 and @xmath242
. these are type assignment statements for @xmath2 , @xmath3 and @xmath4 .
it might be thought that this same technique could be employed for ptss , using type assignment statements for @xmath2 , @xmath3 and @xmath4 , of the form @xmath243 etc and with hypotheses of the form @xmath244 .
this however may not work .
if we have a pts with @xmath245 and can prove @xmath246 and/or @xmath247 , perhaps with @xmath248 , @xmath249 , it may be that ( product ) and ( abstraction ) can not be applied because @xmath250 for any @xmath251 .
this does not mean that @xmath252 can never be cancelled .
we may obtain : @xmath253 where @xmath252 can not be cancelled , as , even if we have @xmath254 @xmath255 and @xmath256 may not be in @xmath23 for any @xmath257 .
however if @xmath258\quad\mbox{,}\quad x\:c\vdash e[y:=q]:s_6 \quad\mbox{and also}\quad(s_1 , s_6 , s_7)\in\mathcal r,\ ] ] so that @xmath259 , we can cancel @xmath252 to give @xmath260).\ ] ] this pts therefore does have theorems not in @xmath19 , but it is hard to determine the hpts corresponding to it .
ptss that have equivalent hptss are @xmath139 and @xmath132coq ( both versions ) , but these belong to a larger class that has the following property : [ d : supers ] a pts is said to be _ supersorted _ if : @xmath261 for supersorted ptss ( abstraction ) can be simplified .
[ t : sixteen ] in every supersorted pts ( abstraction ) can be replaced by : @xmath262 if @xmath263 by lemmas [ l : seven](ii ) and [ l : five ] we have , for some @xmath264 : @xmath265 if the pts is supersorted we have , for some @xmath266 , @xmath267 in the latter ( @xmath268 ) case , and so the result of the former case by lemma [ l : seven](i ) .
hence by ( product ) and supersortedness we have , for some @xmath269 , @xmath270 and by ( abstraction ) we have @xmath271 for a supersorted pts @xmath9 we define a corresponding hpts @xmath52 , which in theorem [ t : twentyeight ] is shown to be equivalent to @xmath52 .
[ d : sshpts ] if @xmath9 is a supersorted pts with specification @xmath272 the _ corresponding _ hpts @xmath52 has specification @xmath273 , with as members of @xmath49 the following theorems of @xmath9 : * axiom to21 pt@xmath274 * @xmath275\:[\pi u{:}s_1.\pi v{:}(\pi x{:}u .
s_2).s_3]$ ] for @xmath276 .
* axiom to21 pt@xmath277 * @xmath278:[\pi x{:}s_1.\pi y{:}x.x]$ ] . *
axiom to21 pt@xmath279 * @xmath280:[\pi x{:}s_1 .\pi y{:}s_2
.\pi z{:}x.\pi u{:}y.x]$ ] .
* axiom to21.6 pt@xmath281 * @xmath282 .
@xmath283\:$ ] @xmath284 @xmath285 $ ] . and additional axioms of @xmath59 generated by ( i ) and ( ii ) : a. if @xmath286 , @xmath287 and @xmath288 is obtained from @xmath32 by replacing any second occurrence of an @xmath158 in @xmath32 by any @xmath159 not in @xmath32 , then if @xmath289 for @xmath290 , @xmath291 .
any conditions on @xmath292 not required in the proof of @xmath289 are not part of the new axiom .
b. if @xmath293\:[\pi x_1{:}a_1\ldots\pi x_{i-1 } { : } a_{i-1}\.s])\in\mathcal b$ ] and @xmath294 satisfies @xmath295 , then @xmath296\ : [ \pi x_1{:}a_1\ldots\pi x_i{:}a_i.s'])\in { \mathcal b}.\ ] ] * note . * the @xmath297 in axioms
i1 , k1 and s1 are sort variables that can be replaced by arbitrary elements of @xmath298 . in the axioms generated by ( i ) and
( ii ) there are restrictions on the sorts that can be substituted for such variables based on the pts provability of the judgements mentioned . given a pts @xmath134x
, we will assume below that @xmath52 is the corresponding hpts .
[ t : seventeen ] if , for a supersorted pts , @xmath299 then @xmath300 . by induction on the derivation of @xmath301 .
if @xmath301 is one of the axioms of @xmath302 , @xmath303 , i1 , k1 or s1 , or is generated by ( i ) or ( ii ) , we have @xmath300 . the ( application ) and ( conversion ) cases follow from the induction hypothesis . the ( subject reduction ) and ( type reduction ) cases follow from the induction hypothesis and lemma [ l : six ] .
[ l : eighteen ] in a hpts corresponding to a supersorted pts , 1 .
if @xmath304 then there is an @xmath305 such that @xmath306 .
if @xmath293\:[\pi x_1{:}a_1\ldots\pi x_{i-1 } { : } a_{i-1}\.s])\in\mathcal b$ ] , there is an @xmath307 such that @xmath308\:[\pi x_1{:}a_1\ldots\pi x_i{:}a_i .s'])\in\mathcal b$ ] .
a. if @xmath309 this follows by supersortedness .
+ if @xmath310 and @xmath311 , we have @xmath300 by theorem [ t : seventeen ] and @xmath31 , for some @xmath106 , by lemma [ l : five ] . + hence @xmath312 by ( i ) .
b. if @xmath313\:[\pi x_1{:}a_1\ldots\pi x_{i-1}{:}a_{i-1}\.s ] ) \in\mathcal b$ ] , by theorem [ t : seventeen ] and lemma [ l : four](iv ) and ( iii ) we have , for some @xmath307 , @xmath314 ( abstraction ) and ( ii ) then give the result .
we also need an extension of @xmath52 that allows hypotheses .
[ d : lam ] if @xmath9 is a pts , @xmath315 has all the postulates of @xmath52 , also with nonempty contexts , and the ( start ) and ( weakening ) rules of @xmath9 .
[ l : nineteen ] @xmath316 immediate because in a derivation of @xmath317 no ( start ) or ( weakening ) rule can be used .
no nonempty context can be emptied in @xmath315 .
the extra axioms of @xmath49 generated by ( i ) we will need in the proof of the correctness of types lemma for @xmath315 ( if @xmath318 then @xmath319 for some @xmath29 ) . those generated by ( ii ) we need in the proof of ( abstraction ) to show that , if we have @xmath320 , we also have for @xmath321 , @xmath322 and @xmath323 , where the derivation of the latter is no longer than the derivation of @xmath320 .
the no longer than " is needed for proof by induction to work . many of the axioms are , in a sense , superfluous .
we can for example , prove axioms @xmath324 and @xmath325 ( below ) from axiom k1 and axiom @xmath326 from axioms @xmath327 and @xmath328 .
however , using fewer axioms can mean that the derivation of a @xmath323 is longer than that of @xmath329 . to illustrate that the axioms , generated by ( i ) and ( ii ) above , form finite sets , we list all the ones generated by axiom @xmath327(@xmath330 below is such that @xmath331 ) .
there are another six i axioms , another sixteen k axioms and twentynine more s axioms .
@xmath332@xmath333:s_4 $ ] .
@xmath334@xmath335\:[\pi u{:}s_1.s_4]$ ] .
@xmath336@xmath337 \:[\pi u{:}s_1.s_2']$ ] . @xmath338@xmath339\:s_3 $ ]
. @xmath340@xmath341\:[\pi u{:}s_1 .\pi
v{:}(\pi x{:}u.s_2 ) .\pi
x{:}u.s_2]$ ] .
@xmath342@xmath343\:s_4 $ ] .
@xmath344@xmath345\:[\pi u{:}s_1.\pi v{:}(\pi x{:}u.s_2).s_3']$ ] .
@xmath346@xmath347\:[\pi u{:}s_1.s_4]$ ] .
@xmath348@xmath349\:[\pi u{:}s_1 .
\pi v{:}(\pi x{:}u.s_2 ) .
s_4]$ ] .
@xmath350@xmath351\:[\pi u{:}s_1.\pi v{:}(\pi x{:}u.s_2).\pi x{:}u .
s_3']$ ] .
the axiom required by ( i ) for @xmath352 and @xmath353 is @xmath354 , for @xmath355 and @xmath356 , for @xmath357 , @xmath358 and for @xmath359 the instance of @xmath360 where @xmath361 .
the axiom required by ( ii ) for @xmath303 is @xmath359 , for @xmath354 is @xmath362 , for @xmath362 is @xmath363 , for @xmath364 is @xmath365 , for @xmath358 is @xmath366 , for @xmath366 is @xmath353 and for @xmath353 is @xmath357 .
each axiom is an axiom scheme in the sense that it is an axiom for all @xmath367 for which it is provable in @xmath9 .
thus most axioms ( not @xmath3686 ) have some restrictions , other than @xmath369 , for example @xmath370 in @xmath371
. some of these restrictions will appear in ( the proofs of ) some of the lemmas for @xmath315 below .
we will show later that in @xmath372 , for a suitable @xmath9 , ( product ) and ( abstraction ) are admissible and that the theorems of @xmath372 are exactly those of @xmath52 and @xmath9 .
to state and prove some preliminary lemmas we need some definitions .
[ d : mpc ] a _ major premise chain ( mpc ) _ in a derivation is a sequence of judgements starting with one formed by a ( start ) rule or an axiom .
the remaining judgements of the chain are obtained by ( weakening ) , ( application ) or ( conversion ) , with the previous judgement as major premise , or by ( subject reduction ) or ( type reduction ) .
the minor premises in ( weakening ) , ( application ) and ( conversion ) rules for which the major premises are in an mpc , will be called the _ minor premises attached to the mpc_. the final judgement of an mpc that is not a proper part of a larger mpc , must be the final judgement in a derivation , a judgement that is the premise for a ( start ) rule or the minor premise in a ( weakening ) , ( application ) or ( conversion ) rule . any derivation is therefore made up of linked mpcs .
[ d : losh ] an mpc is said to be _ long _ if it starts with an axiom of the form @xmath373\ : [ \pi x_1{:}a_1\dots\pi x_n{:}a_n.a_{n+1}]\ ] ] where @xmath374 , @xmath15 is one of @xmath375 or is formed by application from ( some of ) @xmath375 and the mpc has at least @xmath376 ( application ) steps and ( subject reduction ) steps that reduce all of the @xmath376 @xmath377 redexes .
an mpc is _ short _ otherwise .
a derivation is _ short _ if it has no long mpcs and _ long _ otherwise .
[ d : alength ] the _ application length _ or _ alength _ of a derivation is its number of ( application ) steps , where steps in identical minor premises in the derivation , are counted only once . a derivation of lesser alength than another will be called _ ashorter _ , one of greater alength
_ alonger_.
* one derivation of a judgement may be shorter ( in length ) than another without being short .
[ l : twenty ] if the final mpc in a derivation of @xmath378 is long , it starts with an axiom of the form _ ( [ eq : eighteen ] )
_ and the ( subject reduction ) step that reduces the @xmath379 redex comes directly after the @xmath376th ( application ) step , then that derivation of _ ( [ eq : nineteen ] ) _ can be replaced by an ashorter one .
this has to be proved for each of the axioms of @xmath59 that is of this form
. we will prove it for axiom s8 , below , the proofs for other axioms are similar . *
s8 * @xmath380 @xmath381\:$ ] @xmath382 @xmath383 $ ] .
let the minor premises in the six ( application ) steps involving @xmath384 to @xmath385 in the long derivation of ( [ eq : nineteen ] ) be , for @xmath386 : @xmath387 where @xmath388 , @xmath389 , @xmath390 , @xmath391 , @xmath392 , @xmath393 and @xmath394 .
then for some @xmath395 , @xmath396 and @xmath397 and for some @xmath398 $ ] , @xmath399 $ ] , @xmath400.s_3 $ ] and @xmath401 $ ] . note that as contexts can only grow , each @xmath402 for @xmath403 is an initial segment of @xmath404 now by ( weakening ) , ( subject reduction ) , ( type reduction ) and just three ( application ) steps we get from three of these minor premises : @xmath405.s_t,\cr \gamma_6 & \vdash^{h+ } y_5y_6\:r[x_i]\cr \!\hbox to170 pt{and so\hfill}&\hbox to262pt{\hfill}\cr \gamma_6 & \vdash^{h+ } y_3y_6(y_5y_6):s_t\cr } \ ] ] which , as @xmath406 , gives ( [ eq : nineteen ] ) .
we now have a new derivation of ( [ eq : nineteen ] ) , which , given that any ( application)s in the two uses of @xmath407 are counted only once , has fewer ( application)s , and so is ashorter than , the old derivation of ( [ eq : nineteen ] ) .
[ l : twentyone ] every valid judgement in a hpts@xmath408 has a short derivation . we prove this by showing that for every long derivation there is an ashorter derivation of the same judgement .
assume that the following is the part of a long mpc , in a long derivation , up to the @xmath379 reduction , together with the minor premises used in the n ( application ) steps .
@xmath409 @xmath410 @xmath411 @xmath412\ ] ] @xmath413 @xmath414\ ] ] @xmath415 @xmath416\ ] ] @xmath417 @xmath418 @xmath419y_1\dots y_k\:d\vspace{-2mm}\ ] ] here @xmath420 is an axiom of the form ( [ eq : eighteen ] ) with @xmath15 made up of ( some of ) @xmath375 , @xmath421 and @xmath422 the first , second and @xmath376th of the @xmath376 or more ( application)s and the ( subject reduction ) contracting the @xmath379 redex are explicitly shown .
the steps after the @xmath376th ( application ) only alter @xmath423 by reducing it , so steps can be permuted so that the @xmath424 reduction takes place straight after the @xmath376th ( application ) step as follows : @xmath425 @xmath426 @xmath427\vspace{-2mm}\ ] ] @xmath428\:c_n[x_n:=x_n]\vspace{-2mm}\ ] ] @xmath419y_1\dots y_k\:d\vspace{-2mm}\ ] ] this new derivation is no alonger than the original , but the part up to @xmath429\:c_n[x_n:=x_n]$ ] is long and can be replaced , by lemma [ l : twenty ] , by an ashorter derivation , so the whole derivation becomes ashorter .
( if the derivation had identical mpcs to the above , which were all minor premises in the same mpc , all would have to be altered as above to ensure that the new derivation is not alonger than the old . ) in the remaining lemmas and theorems we use a different measure of length of a derivation , where similar " subderivations are counted only once .
[ d : sim ] two derivations are said to be _
similar _ if they are identical or one , in its final mpc , starts with an axiom of @xmath59 of the form ( [ eq : eighteen ] ) , and the other differs only in that its final mpc starts with an axiom of @xmath59 generated from the other by one or more applications of ( ii ) .
we now define : [ d : simlen ] the _ similarity length _ ( or _ slength _ ) of a derivation is given by : a. the number of ( application ) steps , b. the number of ( conversion ) , ( start ) and ( weakening ) steps .
similar derivations ending in the two premises of a ( weakening ) step , are counted only once .
a derivation of lesser slength than another will be said to be _
sshorter _ and one of greater slength as
_ slonger_.
[ l : twentytwo ] given , for @xmath430 , a short derivation of : @xmath431 there is , for some @xmath106 , a derivation , no longer or slonger than that of _ ( [ eq : twenty ] ) _ , of @xmath432 consider the first judgement in the final mpc in a short derivation of ( [ eq : twenty ] ) .
this can not be an axiom of @xmath433 or be formed by a ( start ) rule , so it is an axiom of @xmath434 of the form ( [ eq : eighteen ] ) , where @xmath435 and @xmath436 .
if in this mpc we replace this axiom by the one generated from it by ( ii ) , then using exactly the same steps and minor premises we obtain a derivation of ( [ eq : twentyone ] ) of the same length .
in this final mpc there are no ( weakening ) steps in which the premises are similar , until perhaps after the last ( application ) step , as , until then , no type can be in @xmath298 .
if , at that stage , ( [ eq : twenty ] ) is formed by one or more ( weakening ) steps ( and perhaps ( subject reduction ) ) from @xmath437 and similar minor premises such as @xmath438 , these are counted only once each in the slength . in the derivation obtained by changing the axiom ,
the above derivations remain similar and so the slength of the derivation remains the same .
[ l : twentythree ] if @xmath134x is supersorted and @xmath439 then , for some @xmath106 , @xmath440 or there is a short derivation , of slength no more than that of a short derivation of _ ( [ eq : twentytwo ] ) _ , of @xmath441 by induction on the number k , of judgements in the final mpc of a short derivation of @xmath318 , where @xmath442 .
if k=1 and ( [ eq : twentytwo ] ) comes by a ( start ) rule from @xmath443 where @xmath248 and @xmath444 , ( [ eq : twentythree ] ) comes from two copies of ( [ eq : twentyfour ] ) and ( weakening ) .
the two derivations of ( [ eq : twentyfour ] ) are counted only once , so this derivation of ( [ eq : twentythree ] ) is no slonger than that of ( [ eq : twentytwo ] ) . if ( [ eq : twentytwo ] ) is an axiom we have ( [ eq : twentythree ] ) by ( i ) or by supersortedness .
we now assume k @xmath445 2 .
if ( [ eq : twentytwo ] ) comes from @xmath446 and @xmath447 , by ( weakening ) , where @xmath448 , @xmath449 and @xmath450 , these derivations are both counted in the slength of the derivation of ( [ eq : twentytwo ] ) .
we have , by the induction hypothesis , @xmath451 , by a derivation no slonger than that of @xmath452 , for some @xmath453 and we obtain ( [ eq : twentythree ] ) by ( weakening ) , by a derivation that is no slonger than that of ( [ eq : twentytwo ] ) . if ( [ eq : twentytwo ] ) comes from @xmath454 , by ( subject reduction ) , we have ( [ eq : twentythree ] ) by a derivation no slonger than that of ( [ eq : twentytwo ] ) .
if ( [ eq : twentytwo ] ) comes from @xmath455 , by ( type reduction ) , we have @xmath456 by the induction hypothesis and ( [ eq : twentythree ] ) by ( subject reduction ) , by a derivation no slonger than that of ( [ eq : twentytwo ] ) . if ( [ eq : twentytwo ] ) comes from @xmath455 , by ( conversion )
, we have ( [ eq : twentythree ] ) by a derivation sshorter than that of ( [ eq : twentytwo ] ) . if ( [ eq : twentytwo ] ) comes from @xmath457 and @xmath458 , where @xmath114 and @xmath459 $ ] , by ( application ) , we have by the induction hypothesis , @xmath460 for some @xmath461 , by a derivation no slonger than that of @xmath462 . then by lemma
[ l : twentytwo ] we have @xmath463 , for some @xmath290 by a derivation no slonger than that of @xmath462 . then by ( application ) using @xmath464 we have ( [ eq : twentythree ] ) by a derivation no slonger than that of ( [ eq : twentytwo ] ) . [
l : twentyfour ] if @xmath465 then , for some @xmath106 , @xmath466 by an easy induction on the derivation of ( [ eq : twentyfive ] ) .
[ l : twentyfive ] if @xmath9 is supersorted , ( abstraction ) is admissible in @xmath467 if @xmath9 is supersorted we prove that if @xmath468 then @xmath469 by induction on the slength of a short derivation of ( [ eq : twentysix ] ) . * case 1 .
* ( [ eq : twentysix ] ) comes by ( start ) ( and ( type reduction ) ) from @xmath470 where @xmath248 and @xmath471 . by axiom i1 and ( application ) ( and ( type reduction ) )
we have ( [ eq : twentyseven ] ) .
* case 2 . *
( [ eq : twentysix ] ) comes by ( weakening ) ( and reduction ) from @xmath472 then by the correctness of types lemma ( lemma [ l : twentythree ] ) or supersortedness , for some @xmath473 .
@xmath474 and ( [ eq : twentyseven ] ) follows after three ( applications ) applied to axiom k1 ( and reduction ) .
* ( [ eq : twentysix ] ) comes by ( conversion ) ( and reduction ) from @xmath475 by the induction hypothesis and ( subject reduction ) we have : @xmath476 and @xmath477 where @xmath478 .
we have by lemma [ l : twentythree ] applied to ( [ eq : twentysix ] ) , for some @xmath473 , @xmath479 and , by supersortedness @xmath480 for some @xmath251 , so by axiom @xmath327 and ( subject reduction ) , @xmath481 hence by ( [ eq : twentyeight ] ) and ( conversion ) we have ( [ eq : twentyseven ] ) .
* case 4 * ( [ eq : twentysix ] ) comes by ( application ) ( and reduction ) from @xmath482 and @xmath483 where @xmath484 and @xmath485\to\hskip -7pt \to_{\beta}b$ ] . by the correctness of types
lemma we have for some @xmath486 , by a derivation no slonger than that of ( [ eq : twentynine ] ) : @xmath487 now by lemma [ l : twentytwo ] we have for some @xmath488 , by a derivation no slonger than that of ( [ eq : twentynine ] ) , and so sshorter than that of ( [ eq : twentynine ] ) : @xmath489 now by the induction hypothesis applied to ( [ eq : twentynine ] ) , ( [ eq : thirty ] ) and ( [ eq : thirtytwo ] ) we have : @xmath490 @xmath491 @xmath492 also by the correctness of types lemma applied to ( [ eq : thirtyfour ] ) we have for some @xmath493 @xmath494 and by lemma [ l : twentytwo ] for some @xmath495 @xmath496 now by axiom s1 , @xmath497 , ( obtained as in case 3 ) ( [ eq : thirtysix ] ) , ( [ eq : thirtyfive ] ) , ( [ eq : thirtythree ] ) , ( [ eq : thirtyfour ] ) and five ( application)s , ( subject reduction ) and ( type reduction ) give ( [ eq : twentyseven ] ) .
( note that in axiom s1 @xmath498 and @xmath251 ( here @xmath499 and @xmath266 ) can be arbitrarily chosen in a supersorted pts ) .
[ l : twentysix ] if @xmath9 is supersorted ( product ) is admissible in @xmath500 .
if @xmath501 , @xmath502 and @xmath480 , by lemma [ l : twentyfive ] , @xmath503 so by axiom ii1 , ( application ) and ( subject reduction ) we have @xmath504 lemmas [ l : twentyfive ] and [ l : twentysix ] show that a theorem that can be proved in @xmath505 , using hypotheses , ( abstraction ) and ( product ) , can also be proved in @xmath52 .
so : [ t : twentyseven ] if @xmath9 is supersorted it is equivalent to @xmath315 in that they have the same valid judgements . by theorem [ t : seventeen ] , @xmath52 is a subsystem of @xmath9
. the additional rules of @xmath315 are rules of @xmath9 , so @xmath315 is a subsystem of @xmath9
. the extra rules of @xmath9 have been shown to be admissible in @xmath315 in lemmas [ l : twentyfive ] and [ l : twentysix ] , so @xmath315 and @xmath9 have the same valid judgements .
[ t : twentyeight ] if @xmath9 is supersorted @xmath9 and @xmath52 are equivalent in that they have the same theorems .
it follows from theorem [ t : twentyseven ] that @xmath9 and @xmath315 have the same valid judgements with empty contexts i.e. theorems and so by lemma [ l : nineteen ] that @xmath506 and @xmath52 have the same theorems .
axioms i1 , k1 and s1 can be rewritten in terms of type free combinators ( allowing @xmath507-reduction ) as : * axiom to21 pt@xmath508 * @xmath509 .
* axiom to21 pt@xmath510 * @xmath511 .
* axiom to21 pt@xmath512 * @xmath513 .
@xmath514 $ ] .
these give the standard types of the combinators ( writing @xmath515 for @xmath26 when @xmath516 ) : @xmath517 or , as a special case if @xmath518 , where @xmath519 , were represented as @xmath520 ( as it is in @xmath521 ) , axiom @xmath327 represents the type for @xmath522 ( or @xmath523 ) : @xmath524
in the de bruijn automath systems @xmath368 and @xmath134 are usually identified .
kameraddine has studied the effect of allowing @xmath157-reductions in the ( former ) @xmath368 terms in [ 5 ] .
doing this axiom i1 becomes : @xmath525 and similarly for the other axioms . if we write the type in terms of combinators we can get ( depending on the algorithm ) @xmath526
we have shown that ptss come in at least three categories : those satisfying ( @xmath527 ) and ( @xmath212 ) that have no equivalent hpts , those satisfying ( @xmath528 ) that have only a trivial equivalent hpts and supersorted ptss , such as @xmath67 , that have a nontrivial equivalent hpts .
the standard ptss from the literature that we considered all fit into these categories .
5 barendregt , h.p .
_ lambda calculi with types _ , pp . 117 - 309 in handbook of logic in computer science , vol 2 of oxford science publications , oxford university press , new york , 1992 .
theoretical computer science , 169:3 - 21 , 1996 .
+ bunder , m.w . a deduction theorem for restricted generality .
_ notre dame journal of formal logic _ , 14:341 - 346 , 1973 .
+ bunder , m.w . and dekkers , w.j.m .
equivalences between pure type systems and systems of illative combinatory logics .
_ notre dame journal of formal logic _ , 46:181 - 205 , 2005 .
+ geuvers , h. _ logics and type systems _ , thesis university of nijmegen , 1993 .
+ kamareddine , f. typed @xmath134-calculi with one binder .
_ journal of functional programming _ , 15(5):771 - 796 , 2005 . | for many a natural deduction style logic there is a hilbert - style logic that is equivalent to it in that it has the same theorems ( i.e. valid judgements @xmath0 where @xmath1 . for intuitionistic implicational logic
, the axioms of the equivalent hilbert - style logic can be propositions which are also known as the types of the combinators @xmath2 , @xmath3 and @xmath4 .
natural deduction versions of illative combinatory logics have formulations with axioms that are actual type statements for @xmath5 , @xmath3 and @xmath4 . as pure type systems ( ptss ) are , in a sense , equivalent to systems of illative combinatory logic , it might be thought that hilbert style ptss ( hptss ) could be based in a similar way .
this paper shows that some ptss have very trivial equivalent hptss , with only the axioms as theorems and that for many ptss no equivalent hptss can exist .
most commonly used ptss belong to these two classes . for some ptss however , including @xmath6 and the pts at the basis of the proof assistant coq
, there is a nontrivial equivalent hpts , with axioms that are type statements for @xmath7 and @xmath4 . |
let consider the generic model of inverse problems @xmath0 where @xmath1 and @xmath2 refer to the unknown and the data in a banach space @xmath3 and @xmath4 , and @xmath5 is a nonlinear operator .
the model can represent a variety of inverse problems arising in industrial applications including computerized tomography , inverse scattering and image processing . due to the ill - poisedness of the problem ,
a regularization method must be applied in order to retrieve @xmath6 from the noisy data @xmath7 and there are numerous works devoted for regularization methods .
one of the most appealing regularization techniques is tikhonov regularization , which has been studied from both theoretical and computational aspects by many authors .
the tikhonov regularization takes the minimizer of @xmath8 to the functional @xmath9 in an admissible set @xmath10 : @xmath11 where @xmath12 here @xmath13 is the regularization parameter compromising the data fitting and fidelity term @xmath14 and a priori information encoded in the restoration energy functional @xmath15 of @xmath6 .
commonly used data fidelity functionals @xmath16 include @xmath17 , @xmath18 and @xmath19 and regularization functionals @xmath15 include @xmath20 with a bounded linear self - adjoint nonnegative operator @xmath21 , @xmath22 , etc .
the set @xmath23 which describes _ a priori _ information for the solution @xmath24 is usually set to be weakly closed convex set in @xmath3 . for instance , @xmath25 where @xmath26 is the another banach space whose topology is stronger than that of @xmath3 and @xmath27 is a given constant . for the detail of tikhonov regularization
, we refer to baumeister @xcite , engl , hanke and neubauer @xcite , groetsch @xcite , hofmann @xcite and references therein .
the selection of a parameter @xmath28 is crucial for the stable solution and there is a significant amount of works for the development of methods for choosing a suitable parameter . among others ,
we refer to @xcite .
the minimum value function ( see definition [ def_value ] ) appearing in tikhonov regularization technique is very useful in determining the regularization parameter . in @xcite on several principles such as the generalized principle of discrepancy , the generalized principle of quasisolutions , the generalized principle of smoothing functional , and the principles are investigated by the calculus of the minimum value function .
the value function calculus also gives an insight to well known conventional principles such as morozov principle .
for example , ito and kunisch proposed in @xcite study the morozov principle in terms of the minimum value function for nonlinear inverse problems . on the basis of the value function calculus
, they propose a modified morozov principle . as we see in section 4 , other conventional principles
can also be formulated in terms of the value function . in all the principles
we have mentioned , each of the regularization parameters is determined with an equation including the value function and its higher derivatives .
it can be computationally expensive to solve the equation numerically . in order to reduce the computational effort , an efficient method and an algorithm should be developed .
a model function approach is proposed in @xcite , in which an approximation ( a model function ) to the minimum value function is constructed and use the model function for the value function in their principle .
other principles and numerical algorithms can be found in kunisch and zou @xcite , xie and zou @xcite . in this paper ,
new results on the properties of the minimum value function are derived in a general set - up for problem ( [ tikhonovfunctional ] ) , and a new principle for a choice of a regularization parameter is proposed using the minimum value function , which strongly relates to the principle of reginska @xcite also known as the minimum product criterion .
we also propose a model function for the value function and employ the model function approach in several conventional principles to numerically compute the regularization parameters accurately and efficiently .
the paper is organized as follows . in section 2
, we give properties of the minimum value function .
a model function is proposed in section 3 and the efficiency of the model function is verified in section 4 . in section 5 , a new principle for the regularization parameter is given .
in this section , we investigate the properties of the minimum value function in a general framework : consider the generic model of inverse problems @xmath0 where @xmath5 is a nonlinear operator from a banach space @xmath3 and another banach space @xmath4 .
we retrieve @xmath6 from the noisy data @xmath7 using tikhonov regularization technique , i.e. , @xmath6 is approximately obtained with the minimizer @xmath8 @xmath29 [ def_value ] the minimal value function @xmath30 of @xmath9 is the function of the parameter @xmath28 defined as @xmath31 [ lem : f ] the value minimum function @xmath30 is ( i ) monotonically increasing and ( ii ) concave .
( i ) : let @xmath32 be given .
for any @xmath33 , @xmath34 taking the infimum with respect to @xmath6 yields @xmath35 .
+ ( ii ) : let @xmath36 and @xmath37 be given .
set @xmath38 for @xmath39 , then @xmath40 hence @xmath30 is concave .
since @xmath30 is concave , it is continuous .
[ cor : dfae ] @xmath30 is continuous everywhere .
lemma [ lem : f ] does not require the existence of @xmath33 that achieves the infimum of @xmath9 .
we examine the minimum value function more closely .
let @xmath41 and @xmath42 are one - sided derivatives of the value function @xmath43 , i.e.,@xmath44 note that both limits exist for all @xmath13 and that @xmath45 .
indeed , for given @xmath13 , let @xmath46 .
select @xmath47 as @xmath48 , then @xmath49 . by the convexity of @xmath30 , @xmath50 then @xmath51 therefore , the limit @xmath52 exists . here
we list two the basic properties of @xmath53 . 1 .
* monotonicity * : @xmath54 , for all @xmath32 , 2 . * left and right continuity * : @xmath55 , and @xmath56 for all @xmath13 .
the differentiability of @xmath43 at @xmath28 guarantees the continuity of @xmath57 at this point .
indeed , the monotonicity of @xmath57 and the left continuity of @xmath58 yield the inequalities @xmath59 .
now suppose @xmath43 is differentiable at @xmath28 , i.e , @xmath60 .
then from the inequalities it follows that @xmath61 , which shows the continuity of @xmath41 at @xmath28 .
similarly it follows that @xmath58 is continuous at @xmath28 .
the fact is used in section [ sec : df ] . for the analysis of the minimum value function ,
we introduce the following function of @xmath62 : @xmath63 note that once a result on @xmath30 is proved then the same result is true for @xmath64 as well . for example
, we can show that @xmath64 is concave in exactly the same way as lemma [ lem : f ] .
every results on @xmath64 can be written in terms of @xmath30 by using the relation : @xmath65 which can be verified without difficulty .
we study the asymptotic behavior of @xmath30 .
[ prop : asymp ] @xmath66 let @xmath67 .
for any @xmath68 there exists @xmath69 such that @xmath70 .
then @xmath71 by passing to the limit @xmath72 and taking into account that @xmath73 is arbitrary , we obtain @xmath74 . on the other hand ,
as we see above we have @xmath75 for all @xmath76 , here @xmath28 is arbitrary fixed . since @xmath77 , we obtain @xmath78 which means @xmath79 . therefore @xmath80 . from this inequality
we get @xmath81 .
let @xmath82 be a sequence that converges to 0 as @xmath83 .
we pick @xmath84 such that @xmath85 to get the sequence @xmath86 such that @xmath87 . by monotonicity of @xmath53
we have @xmath88 passing to the limit @xmath89 yields @xmath90 . from this proposition
it follows that @xmath30 is right continuous at @xmath91
. @xmath92 applying proposition [ prop : asymp ] to @xmath64 yields @xmath93 or equivalently , @xmath94 by using the relation ( [ equ : dg ] ) , we arrive at the desired results .
we investigate the relation between the derivative @xmath95 and the two terms , the fidelity functional @xmath15 and the regularization functional @xmath14 .
we also show a sufficient condition for the existence of @xmath95 in terms of @xmath14 and @xmath15 .
we denote the set of solutions of the minimization problem @xmath96 by @xmath97 . for simplicity of our argument
we always assume the existence of the minimizer .
it may happen that there exist two minimizers @xmath98 satisfying @xmath99 and @xmath100 that is , for a fixed @xmath28 , the value @xmath101 and @xmath102 may vary depending on the choice of the minimizer @xmath103 .
thus , it is possible that the maps @xmath104 and @xmath105 will be multi - value functions . in what follows
, we study the basic properties of those functions .
we begin with the following inequality .
[ lem : df ] the following inequalities hold for all @xmath106 : @xmath107 for arbitrary @xmath108 such that @xmath109 , @xmath110 thus , @xmath111
thus we obtain @xmath112 by passing to the limit @xmath113 , it follows that @xmath114 .
similarly , we obtain @xmath115 and thus the inequality ( [ ineq : dfandpsi ] ) is proven .
we also obtain the second inequality ( [ ineq : dfandphi ] ) from ( [ ineq : dfandpsi ] ) and the definition of @xmath30 .
[ cor : df = psi ] if @xmath116 exists at @xmath28 , then @xmath117 and @xmath118 are single valued at @xmath28 and it holds that @xmath119 and @xmath120 for all @xmath121 . note a monotone increasing ( decreasing ) function is differentiable except on a possibly countable set .
[ cor : df = psi ] there exists a possibly countable set @xmath122 such that , * @xmath43 is differentiable and the multi - value functions @xmath101 and @xmath117 have single value on @xmath123 .
* @xmath124 , @xmath125 for all @xmath121 if @xmath123 .
corollary [ cor : df = psi ] guarantees the differentiability of @xmath30 except on a possibly countable set .
next we show the conditions for the differentiability of @xmath30 at all @xmath13 .
firstly , we define the @xmath15 boundness to state the assumptions .
a sequence @xmath126 is @xmath15-bounded if there exists constant @xmath27 such that @xmath127 for all @xmath128 .
[ ass:2 ] * let @xmath129 be a @xmath15-bounded sequence .
there exists a subsequence @xmath130 which weakly converges to an element @xmath131 in the topology of @xmath3 .
* @xmath14 and @xmath15 are lower semi - continuous with respect to weakly convergence sequencers , i.e , if a subsequence @xmath132 which weakly converges to an element @xmath131 , then @xmath133 henceforth hereafter we assume the assumption holds . then is guaranteed the existence of the solutions @xmath134 in @xmath97 such that @xmath135
. there exist @xmath136 and @xmath137 in @xmath97 such that @xmath138 and @xmath139 for all @xmath13 .
let @xmath13 fixed arbitrary and let @xmath140 be a parameter such that @xmath141 and @xmath142 .
let @xmath143 be a minimizing sequence .
then from the monotonicity of @xmath30 , @xmath144 thus @xmath145 and it follows that the sequence @xmath143 is bounded . by assumption [ ass:2 ]
, there exists subsequence of @xmath146 , which we denote it by @xmath146 , that converges weakly to an element @xmath147 .
then by the continuity of @xmath30 and the lower semi - continuity of @xmath14 and @xmath15 , it follows that @xmath148 and thus we have @xmath149 and @xmath150 . in what follows ,
we show that @xmath151 . by lemma [ lem : df ] , @xmath152 the last equality follows from the left continuity of @xmath153 . since @xmath154
, we obtain @xmath155 and therefore @xmath151 .
similarly we can show the existence of the minimizer @xmath156 that satisfies @xmath157 .
we complete the proof .
there exist elements @xmath158 and @xmath159 such that @xmath160 and @xmath161 .
[ cor:22 ] if @xmath162 for all @xmath163 for all @xmath13 , then @xmath116 exists and it is continuous for all @xmath13 . [ cor:2 ] assume that the solution of minimization problem ( [ tikhonovfunctional ] ) is unique for all @xmath13 , then @xmath116 exists and it is continuous for all @xmath13 .
the other properties for @xmath30 such as second differentiability is studied in @xcite . as shown in corollaries [ cor:22 ] and [ cor:2 ] ,
both of the value @xmath30 and @xmath164 is obtained with the knowledge of @xmath8 , and the computation of @xmath165 is not required .
moreover , we obtain @xmath166 and @xmath167 from @xmath168 for @xmath169 provided that @xmath5 is linear from hilbert space @xmath3 to another hilbert space @xmath4 , @xmath170 and @xmath171 with a symmetric operator @xmath21 such that ( [ tikhonovfunctional ] ) has a unique solution , and @xmath172 , i.e , no constraint is imposed .
[ thm : f2k ] the function @xmath30 is infinitely differentiable at every @xmath13 . the derivatives @xmath166 and @xmath173 for each @xmath174 are give with the @xmath175-th derivative @xmath176 as @xmath177 where the constants @xmath178 are recursively defined as @xmath179 with @xmath180 . the proof is based on the following lemma in @xcite .
[ lem : zou ] the function @xmath8 is infinitely differentiable at every @xmath13 and its derivative @xmath181 , for each @xmath174 , is the unique solution to the following equation : @xmath182 from lemma [ lem : zou ] with @xmath183 , one obtains @xmath184 .
suppose @xmath185 with a constant @xmath178 , then lemma [ lem : zou ] yields @xmath186 then we have @xmath187 where @xmath188 is defined as @xmath189 .
by induction the assertion is valid .
in this section , we propose a model function for the value function for linear inverse problems .
firstly , we give a motivation for using the model function approach .
a principle for determining a regularization parameter often requires solving an equation , for example , morozov discrepancy principle takes the parameter @xmath190 that satisfies the equation @xmath191 with noisy data @xmath7 of noise level @xmath192 .
one can apply a newton type iteration to solve the equation , however , the iteration could be numerically expensive .
one strategy to reduce the computational effort is that : first , represent the equation in terms of the value function as @xmath193 .
then , construct a model function @xmath194 to @xmath30 and find the parameter @xmath195 that satisfies the equation @xmath196 .
if the model function approximates to the value function @xmath43 , the parameter @xmath195 thus determined will give a close approximation to @xmath190 .
we assume that @xmath5 is a linear operator form hilbert space @xmath3 and a hilbert space @xmath4 , @xmath197 ( i.e. , no constraint imposed ) , @xmath198 and @xmath199 with a symmetric operator @xmath21 such that ( [ tikhonovfunctional ] ) has a unique solution .
the solution is written as @xmath200 as we see in theorem [ thm : f2k ] , the minimum value function @xmath30 is infinitely differentiable and thus it is reasonable to consider a rational function as a model function , which is briefly mentioned in @xcite .
we propose our model function to @xmath30 of the particular form @xmath201 the derivation of our proposed model function to @xmath30 bases on the following discussion : just for simplicity we assume that @xmath202 and @xmath203 , although our discussion is valid for the infinite dimension framework . the singular value decomposition of @xmath5 yields @xmath204 where @xmath205 ( @xmath206 are the singular values and @xmath207 $ ] and @xmath208 $ ] are the orthogonal matrices , respectively . then @xmath30 is represented as @xmath209 since we assume that @xmath210 is highly ill - conditioned , the singular value @xmath211 decreases rapidly as @xmath175 increases . as a result only the first few @xmath212-terms satisfying @xmath213 in ( [ svdf ] )
will contribute to the sum .
thus we drop off the remaining terms and obtain the approximation @xmath214 the pad approximation to @xmath30 is constructed through the use of several minimizing elements @xmath8 for different values of regularization parameter @xmath28 . for a given interval @xmath215 ,
let @xmath216 are @xmath212 distinct parameters , which we call reference points in the following .
we compute the function values and its derivatives at the reference points to determine @xmath217 unknowns @xmath218 in @xmath194 by the linear system @xmath219 the more accurate model function will be obtained by imposing further conditions on the higher derivatives to @xmath194 , i.e. , we determine @xmath220 unknowns @xmath221 by the system @xmath222 for the solvability of the systems ( [ sym : f ] ) and ( [ sym : f_more ] ) one can refer to @xcite . in the next section ,
we demonstrate that the model function approximates @xmath30 in the interval @xmath223 $ ] .
the differentiation of @xmath164 yields @xmath224 , and thus @xmath225
. therefore it seems that the computation of @xmath226 is required for the evaluation of @xmath227 , however , from theorem [ thm : f2k ] it is enough to solve the equation for @xmath228 for the evaluation of @xmath227 and is not necessary to compute @xmath226 .
in this section , we present numerical tests to illustrate the efficiency of the proposed method using a linear inverse problems `` * heat * '' which is generated from the matlab package developed by hansen @xcite . @xmath14 and @xmath15 are of the forms @xmath229 with @xmath230 matrix @xmath5 and @xmath231 .
we use four conventional principles to determine the regularization parameters : the morozov discrepancy principle , the damped morozov principle , l - curve criterion and the minimum product criterion . + * the discrepancy principle * ( @xcite ) gives the regularization parameter @xmath232 as the solution of the equation @xmath233 * the damped discrepancy principle * ( @xcite ) is a modification of the discrepancy principle that consists in determining @xmath234 such that @xmath235 in this test , we fix @xmath236 .
+ * l - curve criterion * proposed in @xcite gives the regularization parameter @xmath237 which attains the largest magnitude of curvature @xmath238 of the curve @xmath239 the curvature @xmath238 can be written as @xcite @xmath240 + * the minimum product criterion * @xcite takes the regularization parameter @xmath241 as a local minimum point of the function @xmath242 in this test , we fix @xmath236 . + we rewrite all the principles in terms of @xmath43 by using the equations @xmath243 , @xmath244 and @xmath245 to employ model function approach .
our purpose here is to identify the parameters @xmath246 numerically with high accuracy by employing model function approach .
note that we do not attempt to verify the effectiveness of those conventional principles for the inverse problem .
we also note that our approach is not restricted to these principles .
we can apply the approach to any principles that can be formulated in terms of the value function .
although the size of the matrixes @xmath5 for the specific example is fixed to be @xmath247 , our approach is available for much larger scale inverse problems .
the noisy data @xmath7 is generated by adding random noise to the exact data @xmath248 so that @xmath249 the regularization parameter @xmath250 in morozov principle is computed for each noise level as follows : we first compute @xmath251 in ( [ morozov ] ) for 100 uniformly distributed @xmath28-values in the interval @xmath252 $ ] and find apprioximate optimals @xmath108 and @xmath253 that are very close , i.e , @xmath254 and that satisfy @xmath255 .
then a much smaller interval including these parameters is choosen to compute an accurate @xmath256 .
the other parameters @xmath257 in different principles are computed in a similar manner . in each noise level , an approximation to the optimal parameter @xmath256 in morozov principle is computed by employing model functions as follows : firstly , we construct the model function @xmath258 of the form @xmath259 .
the eight unknowns @xmath260 , @xmath261 in this model function are determined by solving the linear system ( [ sym : f ] ) where the value @xmath262 and @xmath263 are obtained at four points @xmath264 . then we solve the equation @xmath265 , which is the model function version of @xmath266 .
we denote the solution by @xmath267 .
we also construct another model function @xmath268 whose sixteen unknowns @xmath269 , @xmath270 are determined by solving the linear system ( [ sym : f ] ) where up to third derivatives @xmath262 , @xmath263 , @xmath271 and @xmath272 at the same points are used .
we denote the solution of the equation @xmath273 by @xmath274 .
the approximated parameters in the other principles are computed in similar manner using the model function @xmath258 and @xmath275 and we denote them by @xmath276 , @xmath277 ( damped morozov for @xmath278 ) , @xmath279 , @xmath280 ( l - curve ) and @xmath281 , @xmath282 ( minimum product for @xmath236 ) , respectively .
all the computed parameters @xmath256 , @xmath283 , @xmath274 etc in each noise level are reported in from table [ table1 ] to table [ table4 ] .
cc .@xmath284 , @xmath276 and @xmath277 .
d - morozov . [ cols="^,^,^,^ " , ] the authors observed that the first derivative @xmath285 did not approximate to @xmath116 . on the other hand @xmath286
was observed to give a very good approximation to @xmath116 .
the equation ( [ dmorozov ] ) is written in terms of @xmath43 as @xmath287 .
this means that the parameter @xmath284 is determined using only @xmath43 .
thus it is enough to give a good approximation to @xmath43 to compute an approximation to @xmath284 . as we expect , the parameters @xmath288 and @xmath289 in damped morozov principle are very good approximations to @xmath284 .
the parameters @xmath283 , @xmath290 and @xmath291 for all nose level are not so accurate .
this is because the equations ( [ morozov ] ) , ( [ l ] ) , ( [ product ] ) contain the first derivative of @xmath43 and @xmath285 does not approximate to @xmath116 . to give a better approximation to @xmath256 , @xmath292 and @xmath293 ,
the model function must approximate to @xmath95 in high accuracy and our model function @xmath275 will be the candidate .
table [ table1 ] and [ table4 ] show that the parameters determined using @xmath275 are very accurate .
on the other hand , @xmath294 are not so accurate , although they are acceptably close to @xmath292 .
an accurate second derivative of @xmath43 is also required for determining the parameter @xmath292 .
figure [ fig : l1_4points.eps ] and [ fig : l3_4points.eps ] show the curvature of the l - curve @xmath238 with its numerical approximations @xmath295 obtained by using @xmath296 and @xmath297 respectively when the error @xmath298 .
the four reference points @xmath299 used to construct @xmath296 and @xmath297 are depicted by bullets ( @xmath300 ) on the curve @xmath238 in each figure .
( solid line @xmath301 ) and @xmath295 ( dashed line @xmath302 ) obtained by using @xmath303 with five reference points @xmath304 @xmath305 . ] ( solid line @xmath301 ) and @xmath295 ( dashed line @xmath302 ) obtained by using @xmath303 with five reference points @xmath304 @xmath305 . ] ( solid line @xmath301 ) and @xmath295 ( dashed line @xmath302 ) obtained by using @xmath303 with five reference points @xmath304 @xmath305 . ]
the approximation @xmath295 ( dashed line @xmath302 ) in figure [ fig : l1_4points.eps ] completely fails to approximate to @xmath238 .
we used @xmath262 and @xmath263 for the construction of the model function @xmath306 , and thus the second derivative @xmath307 can not approximate to @xmath308 which is contained in @xmath238 . on the other hand ,
our @xmath275 gives better approximation to @xmath309 as shown in figure [ fig : l3_4points.eps ] , although @xmath295 does not much perfectly with @xmath309 . to give more accurate @xmath295
, we construct another model function of the form @xmath310 using @xmath262 , @xmath263 , @xmath271 and @xmath272 at five reference points @xmath311 .
figure [ fig : l3_5points.eps ] depicts the curvature of the l - curve @xmath238 with its numerical approximations @xmath295 obtained by using @xmath303 .
the five reference points are also shown in the figure .
we observe that the model function @xmath303 yields the sufficiently good approximation @xmath295 that almost perfectly matches with the exact curvature @xmath238 .
this observation suggests that we should use more reference points to construct model function when we employ a principle that contains second derivatives of the value function . in a practical situation ,
an interval where a regularization parameter to be found is often much smaller than the interval @xmath312 used for our numerical test .
if it is the case it is enough to construct a model function in a smaller interval .
the number of reference points to be used for the construction of a model function can be reduced to two or three .
we propose a new criterion for the regularization parameter .
let us introduce a function @xmath313 defined as @xmath314 where @xmath315 is a positive constant .
our new criterion takes the parameter as a local minimum @xmath28 of @xmath316 .
since @xmath317 and @xmath318 for all @xmath13 , @xmath319 solves the equation @xmath320 .
the criterion is similar to the minimum product criterion by regiska @xcite .
first , we note that the energy function @xmath321 in ( [ product ] ) is written in terms of @xmath30 @xmath322 suppose that @xmath323 exists and @xmath324 .
( a sufficient condition for the existence of the second derivative and the negativity can be found in @xcite . ) since @xmath325 , the regularization parameter determined by the criterion solves the equation @xmath320 .
the relationship between ( [ newf ] ) and ( [ minimumf ] ) follows from the next proposition .
let @xmath62 be a positive number .
@xmath326 the equality holds if and only if @xmath28 solves the equation @xmath327 .
consider the inequality @xmath328 with @xmath329 and @xmath330 .
substituting @xmath331 with @xmath332 and @xmath333 with @xmath334 , it follows that @xmath335 and the inequality holds if and only if @xmath336 , namely , @xmath337 multiplying @xmath338 and taking @xmath339 power yields the desired inequality .
figure [ fig : gamma.eps ] shows @xmath340 and @xmath341 in the interval @xmath342 for certain linear inverse problem .
( solid line @xmath301 ) and @xmath341 ( dashed line @xmath302),title="fig : " ] + there exists a local minimum point around @xmath343 where @xmath340 and @xmath341 take the same value . the advantages of our criterion are ( i ) the shape of @xmath313 is sharper than @xmath321 and thus it is easier to detect the minimum point .
( ii ) @xmath313 contains @xmath30 only , does not require the function @xmath116 which can be discontinuous due to the nonuniqueness of an inverse problem .
the effect of the parameter @xmath315 in @xmath344 to the quality of the solution @xmath8 should be studied .
we investigate both the applicability of the criterion to nonlinear problems and the effect of the parameter @xmath315 in our future works .
we investigate the minimum value function for the tikhonov regularization .
we propose the model function for the minimum value function for linear inverse problems and verify its efficiency in the determination of the regularization parameter .
we also propose a new criterion for the choice of the regularization parameter .
our criterion strongly relates to the minimum product criterion and is applicable to nonlinear inverse problems . c. r. vogel , non - convergence of the @xmath345-curve regularization parameter selection method , , 12 , 535547 , 1996 . j. xie and j. zou , an improved model function method for choosing regularization parameters in linear inverse problems , , 18 , 631643 , 2002 | the minimum value function appearing in tikhonov regularization technique is very useful in determining the regularization parameter , both theoretically and numerically . in this paper
, we discuss the properties of the minimum value function .
we also propose an efficient method to determine the regularization parameter . a new criterion for the determination of the regularization parameter is also discussed . |
20 killian tc , kulin s , bergeson sd , orozco la , orzel c and rolston sl 1999 4776 | recent experiments with ultracold neutral plasmas show an intrinsic heating effect based on the development of spatial correlations . we investigate whether this effect can be reversed , so that imposing strong spatial correlations could in fact lead to cooling of the ions .
we find that cooling is indeed possible .
it requires , however , a very precise preparation of the initial state .
quantum mechanical zero - point motion sets a lower limit for ion cooling .
experimentally , ultracold ( @xmath0 k ) neutral plasmas could be realized only recently by photoionizing a cloud of ultracold atoms @xcite . from a plasma physics perspective , they are very appealing since they might provide laboratory realizations of so - called strongly coupled plasmas , where the coulomb interaction ( inversely proportional to the mean inter - particle spacing @xmath1 ) dominates the random thermal motion ( proportional to the temperature @xmath2 ) . as a consequence ,
interesting ordering phenomena such as coulomb crystallization can be observed . in order to be in this strong - coupling regime ,
a system has to be either very dense ( large coulomb interaction ) or very cold ( little thermal energy ) .
the first scenario is realized in some astrophysical contexts , which are difficult to access experimentally .
hence , the possibility to realize a strongly coupled plasma in the laboratory by going to ultralow temperatures offered exciting prospects within the plasma physics community . in @xcite ,
an ultracold neutral plasma has been created by photoionizing a gas of laser - cooled xe atoms .
the extremely low temperatures of the xe gas suggested that the system was well within the strong - coupling regime , the so - called _ coulomb coupling parameter _
@xmath3 , i.e. the ratio of potential to thermal energy , being of the order of @xmath4 for the electrons and even up to @xmath5 for the ions . however , it was realized quickly @xcite that despite the low temperature of the xe gas a strongly coupled plasma could not be created in this way . due to the fact that the neutral xe atoms initially present interact only very weakly , the plasma is created in a completely uncorrelated non - equilibrium state .
hence , the subsequent conversion of potential into kinetic energy rapidly heats both the electron and ion subsystem , suppressing the development of substantial correlations @xcite .
this effect has been termed `` disorder - induced heating '' @xcite or `` correlation - induced heating '' @xcite , expressing the fact that it is the development of spatial correlations during the relaxation of the plasma towards an equilibrium state which leads to an increase in temperature . put another way , in the initial non - equilibrium state the potential energy is higher than in the corresponding equilibrium state of a plasma of the same average density .
consequently , relaxation towards thermodynamic equilibrium will reduce the potential energy in the system and increase the kinetic energy , i.e. the temperature .
additionally , the electrons are heated by three - body recombination as well as continuum lowering . suppression of electron heating is thus much more difficult than simply reducing the correlation heating . at present
, there is a proposal of mixing the plasma with a gas of rydberg atoms , so that collisional ionization of these rydberg atoms removes energy from the electron component of the plasma @xcite . for the creation of a strongly coupled ionic component , different procedures
have been suggested to avoid these heating effects , namely ( _ i _ ) using fermionic atoms cooled below the fermi temperature in the initial state , so that the fermi hole around each atom prevents the occurence of small interatomic distances @xcite ; ( _ ii _ ) an intermediate step of exciting atoms into high rydberg states , so that the interatomic spacing is at least twice the radius of the corresponding rydberg state @xcite ; and ( _ iii _ ) the continuous laser - cooling of the plasma ions after their initial creation , so that the correlation heating is counterbalanced by the external cooling @xcite .
so far , none of these procedures has been realized experimentally , however it has been shown theoretically that continuous laser - cooling can indeed lead to a strongly coupled plasma , connected with crystallization of the expanding system @xcite . in this case , the correlation heating effect is still present and is offset by an additional external cooling mechanism . proposals ( _ i _ ) and ( _ ii _ ) , on the other hand , aim at suppressing the heating directly by avoiding the small interparticle distances leading to a large potential energy , i.e. by introducing ( spatial ) correlation in the initial state . as has already been noted in @xcite , this
could also be achieved using optical lattices to arrange the atoms . at this point ,
the question arises naturally whether the correlation heating could not only be avoided , but whether it can actually be reversed and turned into cooling .
as argued above , correlation heating describes the fact that the uncorrelated ions created in the photoionization process have a higher potential energy than in the thermodynamical equilibrium state .
hence , equilibration will reduce the potential energy and increase the kinetic energy of the system , making the ions hotter than the atoms in the initial gas state .
on the other hand , it is conceivable that the atomic gas might be prepared in an `` overcorrelated '' state , where the potential energy is lower than its equilibrium value . in this case
, equilibration must increase the potential energy on the cost of the kinetic energy , i.e. the temperature must _
decrease_. this effect of `` correlation cooling '' has been described before in @xcite and , based on a kinetic approach , in more general terms in @xcite . in @xcite ,
the overcorrelated initial state was achieved by suddenly decreasing the debye screening length , i.e. the temperature of the electrons , rather than by inducing spatial order .
it was shown that in this way , a reduction in temperature of the order of a few percent does indeed occur for the temperatures and densities considered . in the present letter
, we explore an alternative scheme for achieving correlation cooling , namely the use of an optical lattice to induce spatial correlations in the initial state as suggested in @xcite .
we describe , both analytically and by molecular dynamics simulations , the situation of a fully ordered initial state where the atoms have been arranged in a perfect bcc - type lattice structure using an optical lattice . in reality
, their spatial distribution will be broadened , classically due to their non - vanishing kinetic energy corresponding to the initial - state temperature , quantum mechanically corresponding to the eigenstates of the harmonic oscillator potential generated by the optical lattice .
this broadening is neglected in our numerical simulations , however , its influence will be discussed in the analytical considerations below . for the sake of numerical simplicity
, our initial state corresponds to a sphere cut out from a bcc lattice rather than a more realistic cylindrical shape , arguing that edge effects should be of minor importance if the number of atoms in the system is large enough .
the system is propagated using the hybrid molecular dynamics method described in @xcite .
briefly , an adiabatic approximation is employed for the electrons , and their distribution is obtained from the steady - state king distribution @xcite , widely used in the study of globular clusters . the ions are then propagated using the electronic mean - field obtained from the steady - state distribution and by fully taking into account the ion - ion interactions . in the experiments under consideration , quasineutral plasmas with @xmath6 as low as @xmath7 @xcite have been realized , making debye screening effects negligible , while ion numbers @xmath8 up to @xmath9 have been obtained @xcite . for our simulations , such a large particle number would lead to a prohibitively large numerical effort .
however , the typical timescale for the plasma expansion is @xmath10 , where @xmath11 is the einstein frequency in the wigner - seitz model of a coulomb solid and @xmath12 is the ion mass . on the other hand ,
the relaxation of the ions takes place on a timescale of @xmath13 .
hence , the same macroscopic behavior can be obtained by decreasing @xmath8 and increasing @xmath6 , keeping the ratio @xmath14 fixed .
therefore , for our simulations we chose @xmath15 and an initial electron energy corresponding to @xmath16 , while the bare coulomb potential was used for the ion - ion interaction . figure [ fig1](a ) shows the time evolution of the average ion kinetic energy in units of the initial - state temperature for a typical realization .
since the initial lattice configuration produces an effective oscillator potential for each ion , the relaxation process is connected with transient oscillations of the kinetic energy which are damped out due to ion - ion collisions . additionally , the average kinetic energy increases at later times due to a radial ion acceleration caused by the thermal electron pressure , which is proportional to the density gradient .
therefore the slow expansion of the unconfined plasma leads to a softening of the initially sharp plasma edge and a modification of the velocity distribution in the outer region of the plasma only .
since this distribution is close to a maxwellian in the central region of the plasma , the average kinetic energy is directly related to the ionic temperature @xmath2 , while it becomes more and more dominated by the expansion energy towards the edge of the plasma . in order to highlight these edge effects ,
we have calculated the average ion kinetic energy from the velocity distribution in the inner region only , taken to be a sphere with half of the initial plasma radius .
the resulting curve is shown as the dashed line in figure [ fig1](a ) .
clearly , the expansion hardly affects the kinetic energy in the inner region of the plasma , leaving only the oscillations which are damped out as the system approaches thermodynamic equilibrium . as discussed above
, the relaxation of the ions towards equilibrium takes place on a timescale of @xmath17 , while the typical timescale for the plasma expansion is @xmath18 .
hence , a sufficiently large number of ions allows for an almost unperturbed relaxation in the inner plasma region so that a temperature can be defined in a meaningful way .
( for a strontium plasma with a typical density of @xmath19@xmath20 used , e.g. , in @xcite , @xmath21s , while @xmath22 is about @xmath23 times larger for @xmath24 and @xmath25 . ) using the absorption imaging techniques described in @xcite it should be possible experimentally to probe this inner region only .
therefore , we chose to define the final temperature as given by the inner plasma region and to neglect the finite - size effects associated with the boundary region of the plasma .
the resulting @xmath26 was estimated from the center of the remaining oscillations and is shown in figure [ fig1](b ) as a function of @xmath27 by the full circles .
it should be noted that the value of @xmath27 has no meaning as a measure of nonideality of the initial plasma state , since the system may be far from equilibrium .
therefore , in the present case it should simply be viewed as a density - scaled inverse temperature rather than a coulomb coupling parameter . as can be seen in the figure , the ratio of final to initial temperature reaches a constant value of about 0.5 as the initial temperature goes to zero .
hence , cooling can indeed be achieved on a moderate level of a temperature reduction of about 50 percent . for large particle numbers and
if @xmath28 is small enough such that three - body recombination and debye screening can be neglected , quasineutrality leads to a homogeneous electron background in the central plasma region .
thus , we can use the one - component plasma model to gain more insight into the cooling process .
this model has been studied extensively over the last decades @xcite , which led to a detailed understanding of the equation of state over a wide parameter range . in the classical regime ,
the energy per particle of the final equilibrium state is given by @xmath29 where the excess energy @xmath30 accounts for nonideal effects due to the inter - particle coulomb interaction .
for this quantity , several fit formulae can be found in the literature , which have been obtained by monte carlo and molecular dynamics methods @xcite . in the fluid phase we use the interpolation formula from @xcite @xmath31 where @xmath32 , @xmath33 and @xmath34 .
equation ( [ eq2 ] ) satisfies the known small-@xmath35 limit @xcite and accurately fits the high-@xmath35 values obtained by numerical calculations @xcite .
in the solid phase ( @xmath36 ) , the excess energy can be described by an expansion in powers of @xmath37 @xcite @xmath38 where @xmath39 , @xmath40 and @xmath41 . here
, the first term corresponds to the madelung energy of the lattice , with @xmath42 for a bcc - type lattice .
the second term describes thermal contributions corresponding to an ideal gas of phonons while the higher - order terms account for anharmonic corrections .
assuming an unchanged ionic phase - space distribution function directly after the photoionization , where the spatial distribution due to the finite temperature is represented by gaussian profiles centered on the sites of the bcc lattice , the internal energy of the initial nonequilibrium state is given by @xmath43 \ ; , \ ] ] where @xmath44 , with @xmath45 for a bcc lattice . while the first two terms correspond to the ion kinetic energy and the madelung energy of a bcc lattice of point charges , the last two terms give the leading - order corrections due to a broadened charge distribution at the lattice sites expanded in terms of the mean - squared ion displacement @xmath46 . assuming an initially non - interacting atomic gas and harmonic trapping potentials we have @xmath47 where @xmath48 is the oscillation frequency of the trapping potentials .
the final temperature is now simply obtained from energy conservation @xmath49 with @xmath50 for @xmath51 and @xmath52 for @xmath53 .
figure [ fig1](b ) shows the ratio @xmath54 as a function of @xmath27 for @xmath55 , i.e. for a point - like ionic density at the lattice sites .
a comparison with our numerical data shows that equation ( [ eq6 ] ) reproduces the numerical values for a finite - size plasma quite well .
the remaining discrepancies should be attributable to finite - size effects @xcite . at small values of @xmath27
the temperature ratio starts out from unity and decreases with increasing @xmath27 , since the excess potential energy of the final fluid state is naturally larger than that of the initial lattice configuration .
as @xmath27 exceeds a value of @xmath56 the temperature ratio rises linearly , connected with a solidification of the plasma , where @xmath57 stays constant ( see inset ) . in this regime ,
regions of fluid and solid phases coexist in the plasma . at @xmath58
the system has crossed the melting point and the temperature ratio approaches a value of @xmath59 .
this factor of @xmath59 originates from the additional thermal energy of @xmath60 arising from thermal lattice excitations .
however , the situation changes if a finite extension of the initial charge distribution at the lattice sites is considered . neglecting terms of order @xmath61 in equation ( [ eq4 ] )
, one finds that for finite @xmath48 the temperature approaches a value of @xmath62 if @xmath63 .
let us now turn to the question of the influence of possible lattice defects , i.e. how perfect must the initial - state lattice be in order to observe correlation cooling ?
such defects are easily accounted for in the previous consideration . introducing a filling factor @xmath64 , defined as the probability that a given lattice site is occupied
, simply reduces the average charge at each lattice site .
therefore , the potential energy of the initial state can be written as @xmath65 , where @xmath66 corresponds to the temperature scaled by the density of the perfectly filled lattice , which is related to the actual value of @xmath27 by @xmath67 . substituting this in equation ( [ eq4 ] )
yields @xmath68 as the generalization of equation ( [ eq6 ] ) to non - perfect filling . in figure [ fig2](a ) , we show the resulting final temperature as a function of the filling factor for two different @xmath27 . a decreasing filling enhances the initial potential energy ( compared to that of a full lattice with the same average _ density _ , _ not _ the same _ lattice constant _ ) , leading to a rapid increase of the final temperature . as a consequence , cooling can be achieved for a nearly perfect lattice only . furthermore , for a fixed @xmath69 , increasing @xmath27 leads to larger values for the temperature ratio .
this heating also influences the possibility of the creation of a crystallized plasma phase , since the melting point of @xmath57 is shifted to larger values of @xmath27 , as shown in the phase diagram figure [ fig2](b ) . for values of @xmath70 , the system always ends up in the fluid phase independent of @xmath27 .
the above picture suggests that , for a perfect initial state , cooling is always possible independent of the initial temperature . at low enough temperatures , however ,
the classical treatment will break down , and quantum mechanical effects will become important . as in the classical case
the quantum statistical internal energy of a coulomb solid is of the order of 10000 , i.e. well within the crystallized regime . ]
can be written as a sum of the madelung energy , harmonic contributions arising from linear lattice excitations and anharmonic corrections .
for the harmonic contributions to the excess energy we use an analytic approximation formula obtained recently @xcite , which reproduces the numerical results for harmonic coulomb crystals very accurately .
the anharmonic corrections are taken from @xcite , where they were obtained by fitting the results of quantum monte carlo simulations .
the resulting expression is too lengthy to be reproduced here , but incorporates both the known semiclassical high - temperature limit as well as the quantum mechanical groundstate energy of a coulomb crystal , @xmath71 here , @xmath72 measures the importance of quantum effects , i.e. zero - point ion oscillations , and @xmath73 is the wigner - seitz radius in units of the ionic bohr radius . as in the previous classical considerations , we apply a sudden approximation assuming an unchanged density matrix directly after the photoionization pulse .
the initial energy of the plasma is thus found from the expectation value of the new hamiltonian using a density matrix representing independent harmonic oscillators arranged on a bcc lattice .
the result is @xmath74\ ; , \ ] ] which exactly corresponds to the classical expression equation ( [ eq4 ] ) except that the kinetic energy @xmath75 is replaced by its quantum mechanical counterpart .
moreover , the mean - squared displacement is now given by @xmath76 rather than equation ( [ eq5 ] ) , where @xmath77 .
the final temperature is again obtained by equating the initial and final energies of the plasma .
figure [ fig3 ] shows @xmath78 as a function of @xmath27 for @xmath79 .
in contrast to the classical case , here the final temperature can not be decreased to arbitrarily low values , rather it approaches a finite constant value as @xmath80 , since the initial charge distributions at the lattice sites have a finite extension at @xmath81 due to zero - point oscillations . using the low - temperature limit equation ( [ eq8 ] ) as an estimate for the final energy and neglecting terms of order @xmath82 in the resulting energy balance , this limiting temperature is found to be @xmath83 where @xmath84 . since @xmath85 independent of @xmath86 ,
the limiting temperature is always positive and takes on a minimum value of @xmath87 at an optimum ratio of @xmath48 and @xmath88 of @xmath89 note , however , that the limiting temperature obtained numerically with the full expression for @xmath90 is slightly smaller since at finite temperature the system will not be in the groundstate , hence its potential energy is somewhat larger than that given by equation ( [ eq8 ] ) .
we may rewrite equation ( [ eq12 ] ) as @xmath91}\stackrel{r_{\rm s } \gg1}{\longrightarrow}7.424r_{\rm s}^{-1/4}$ ] , where @xmath92 is the thermal debroglie wavelength .
this would suggest that it is possible to reach the degenerate regime of @xmath93 if @xmath86 can be decreased sufficiently , i.e. to a value @xmath94 .
even for hydrogen this would correspond to an unrealistically large density of @xmath95@xmath96 .
note , however , that the value of @xmath86 enters the ratio @xmath97 with a power of @xmath98 only , so that values of @xmath99 can already be reached if @xmath100 , corresponding to a density of @xmath101@xmath96 , which has already been surpassed with cold bose - condensed hydrogen atoms @xcite . in this regime ,
the present considerations will not hold , since effects of particle statistics will be important which have not been included in the above expressions for the energies . however , equation ( [ eq12 ] ) might serve as an estimate for the parameter values necessary to reach the degenerate regime . in summary
, we have discussed the possibility of correlation cooling and of the creation of a strongly coupled ultracold neutral plasma by photoionization of a highly ordered state of cold atoms , achievable with optical lattices .
classical considerations have shown that the initial temperature can be reduced by a factor of two if the atomic sample is noninteracting initially , offering the possibility to reach ion temperatures which have not been achieved so far .
the quantum regime has been investigated using existing expressions for the equilibrium internal energy of the system .
here it was found that the final temperature approaches a constant value which arises from zero - point oscillations in the lattice potentials and depends on the density as well as the geometry of the optical lattice .
an almost perfect lattice is necessary to observe the discussed cooling effects , since even a small amount of lattice defects largely increases the final temperature .
however , it has been demonstrated that it is possible to create regular structures with a well - defined number of atoms localized in the potential wells of an optical lattice @xcite . financial support from the dfg through grant ro1157/4 is gratefully acknowledged . |
the measurement of mass - loss rates for f m dwarf stars and the development of theoretical models to predict these rates have challenged observers and theoreticians for many years . while _ in situ _ measurements of the solar - wind mass flux have been available since the early space age , such measurements are not feasible for other stars .
instead , @xcite showed that astrospheric absorption in the stellar lyman-@xmath0 emission line provides a sensitive tool for measuring stellar mass - loss rates of late - type dwarf stars .
although other techniques have been proposed , including the search for charge - exchange - induced x - ray emission @xcite and free - free emission at radio wavelengths @xcite , the lyman-@xmath0 astrospheres technique is presently the only technique that provides credible results .
we describe this technique in section 2 , present an alternative way for explaining the empirical relation of mass loss to x - ray flux in section 3 , and identify some critical issues in section 4 .
the interaction of ionized stellar - wind plasma with inflowing , mostly neutral , interstellar gas leads to a rich phenomenology .
figure 1 shows an example of a hydrodynamic model of the heliosphere computed by @xcite .
the salient features of this model and other models of the heliosphere and astrospheres is the presence of a termination shock where the solar wind becomes subsonic , a heliopause / astropause that separates ionized solar / stellar wind plasma from charge - exchanged local interstellar medium ( lism ) ions , and the so - called hydrogen wall where inflowing neutral hydrogen gas piles up , is decelerated , and is heated by the charge - exchange reactions .
ibex results @xcite indicate that the inflow velocity of the lism is likely too slow to produce a bow shock , however @xcite argue that inclusion of inflowing he@xmath1 ions yields alfvn and fast magnetosonic speeds slower than the inflow speed and thus a bow shock . as first predicted by @xcite ,
hydrogen walls provide the observational basis for measuring stellar mass - loss rates .
however , the small hydrogen column densities in these astrosphere regions , @xmath214.015.0 , means that optical depths in neutral metal lines also formed in these regions will be too small to detect , leaving lyman-@xmath0 as the only available spectral diagnostic for measuring stellar mass - loss rates .
figure 2 shows the effect of neutral hydrogen absorption from the lism and in the hydrogen walls of the star and the sun on the lyman-@xmath0 emission line emitted by a star .
since the hydrogen wall gas is decelerated relative to the lism gas , the solar hydrogen wall appears as extra absorption on the red side of the lism absorption , and the stellar hydrogen wall viewed from outside of the star appears as extra absorption on the blue side of the lism absorption .
figure 3 shows the effects of increasing stellar mass - loss rate on the stellar hydrogen - wall absorption computed with hydrodynamic astrosphere models @xcite .
this technique has been successfully applied to 15 stars or binary systems that are relatively close ( @xmath3 pc ) , with relatively small lism neutral hydrogen column densities [ @xmath4 .
clearly , the lism gas must contain a significant amount of neutral hydrogen for the development of a hydrogen wall and for this technique to work .
lyman-@xmath0 observations of more stars are needed , but observational limitations will restrict the number of stellar winds that could be measured by this technique .
@xcite proposed that the mass - loss rate per unit stellar surface area ( @xmath5 ) measurements for g and k dwarf stars shown in fig .
4 can be fit by a power law , @xmath6 , for x - ray surface fluxes @xmath7 erg @xmath8 s@xmath9 .
while the fit looks good considering the roughly factor of two uncertainties in the mass - loss rate measurements and the varying x - ray fluxs that were measured at different times , there is no accepted explanation for this power - law relation .
we therefore propose an alternative way of characterizing these data that suggests a physical scenario that could explain these results .
we start by separating the data into three groups rather than by assuming a power - law relation .
group 1 includes the sun and four other stars ( 61 vir , the unresolved binary @xmath0 cen ab , @xmath10 ind , and 61 cyg a ) .
all of these stars are slow rotators ( 2242 day periods ) , most have solar - like mass - loss rates ( per unit surface area ) , @xmath11 , and @xmath12 in the range @xmath13@xmath14 ergs @xmath8 s@xmath9 .
@xcite , @xcite , and others have shown that the solar mass - loss rate is uncorrelated with @xmath12 , although both @xmath5 and @xmath12 vary by over a factor of 10 . the average solar mass loss rate ( @xmath15 yr@xmath9 ) , however , is independent of @xmath12 despite the very different number of active regions on the solar disk at times of solar activity minima and maxima .
group 2 consists of two stars and two unresolved binary systems with @xmath12 in the narrow range 35@xmath16 ergs @xmath8 s@xmath9 .
the two stars ( @xmath10 eri and @xmath17 boo b ) are moderate speed rotators ( @xmath18=11.68 and 11.5 days , respectively ) .
the other two targets are unresolved binaries ( 70 oph ab and 36 oph ab ) with rotational periods in the range 19.722.9 days and measured mass - loss rates only for the binary systems rather than for the individual stars .
the third group of g k dwarf stars consists of @xmath19 uma @xcite and @xmath17 boo a @xcite .
these stars are rapid rotators ( @xmath18 = 4.69 and 6.2 days , respectively ) , have large @xmath20 ergs @xmath8 s@xmath9 , but small mass - loss rates similar to the sun .
we note that these mass - loss rates are a factor of 100 times smaller than theoretical estimates for @xmath19 uma @xcite and for @xmath21 boo , an f7 v star with @xmath22 days @xcite . the very different behavior between mass loss , which emerges only along open magnetic - field lines , and x - ray flux , which is emitted mostly in closed magnetic - field regions with high electron densities , likely depends on the energy available for driving mass loss at the coronal base of open field lines , the number of open field lines , and the divergence of the open field lines with height in the corona .
while detailed theoretical studies of wind acceleration mechanisms ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and coronal magnetic - field morphologies ( e.g. , * ? ? ?
* ) are needed , we outline here a simple physical model that may explain the complex relationship between @xmath5 and @xmath12 shown in fig . 4 .
@xcite showed that for the sun , the mass flux along an open magnetic - field line is proportional to the magnetic - field strength at its coronal base footpoint . since the coronal base magnetic - field strength near active regions is much larger than for high - latitude coronal holes , the mass flux in field lines located near active regions is much larger than for open field regions away from active regions ( e.g. , polar holes ) .
however , the divergence of field lines emerging from near active regions is much larger than for open field regions away from active regions . as a result ,
the greater divergence of field lines from active regions compensates for the larger mass flux in these field lines leading to the same average solar mass - loss rate even when the surface coverage of active regions changes as indicated by a factor of 10 range in @xmath12 .
this explanation for the nearly constant value of @xmath5 over at least a factor of 10 range in @xmath12 seen in the solar data likely also explains the results for the other group 1 stars .
for the faster rotating stars in group 2 , especially @xmath10 eri and @xmath17 boo b , the much stronger magnetic dynamo produces much larger magnetic fluxes @xcite .
this likely means that more active regions are distributed across their stellar surfaces and stronger magnetic - field strengths outside of these active regions are present in the group 2 stars compared with the group 1 stars . as a result , the coronal surface area not covered by active regions is decreased substantially such that there is much less available volume for open field lines to expand .
the smaller divergence of field lines from active regions will not be able to compensate for the enhanced mass flux along these stronger magnetic - field lines .
@xcite showed that in the absence of significant magnetic - field divergence , stellar mass - loss rates will increase with the magnetic field strength and the available energy at the base of the magnetic field lines .
a viable explanation for the solar - like mass - loss rates for the rapidly rotating stars in group 3 is a matter of speculation at this time .
we suggest that the very strong dynamos in these stars produce complex magnetic - field morphologies including a strong toroidal component seen in rapidly rotating stars like @xmath17 boo a @xcite .
these very strong magnetic fields are likely closed in very active stellar coronae leaving only a few open - field lines from which the stellar wind can emerge .
the fewer open - field lines could more than compensate for the stronger mass - flux rates in the remaining open - field lines .
simulations by @xcite of mhd wave propagation upwards along open magnetic field lines from the photospheres of solar - like stars may explain the empirical correlation of @xmath5 with @xmath12 shown in figure 4 .
they found that with increasing magnetic field strength and turbulence in the photosphere ( the origin of stellar activity ) , the reflection of upwardly propagating alfvn waves is less efficient , leading to an increase in the energy available in the corona for mass loss and x - ray emission .
however , radiative losses in the chromosphere increase even faster than the mass - loss flux , leading first to @xmath5 saturation at a high level and then to decreasing @xmath5 with increasing @xmath12 .
qualitatively , this may explain the increase and then decrease of @xmath5 with increasing @xmath12 , but the peak @xmath5 corresponding to saturation in these calculations is much larger than is observed and the decrease in @xmath5 at very large @xmath12 levels was not modeled in detail .
the study of cool dwarf star astrospheres has made important progress from the early theoretical models of the heliosphere that first showed the presence of a hydrogen wall @xcite , through the development of sophisticated - hydrodynamic models of stellar atmospheres by mller , izmodenov , zank , and others , to the testing and refinement of these models by comparison with stellar lyman-@xmath0 observations . despite this important progress ,
there remain a number of critical problems requiring future theoretical and observational studies : * charge - exchange rates * : : the existence and physical properties of solar / stellar hydrogen walls depend critically on charge - exchange reaction rates between outflowing solar / stellar wind ions ( primarily protons ) and the inflowing lism neutral hydrogen atoms .
if the presently assumed charge - exchange reactions rates are in error , then the inferred values for n(h i ) , heating , and deceleration of neutral hydrogen atoms in the hydrogen wall will also be in error .
it is important to compute charge - exchange reaction rates for the low densities and likely non - maxwellian velocity distributions of the interacting species .
* proper inclusion of neutrals * : : neutral hydrogen and helium atoms inflowing from the lism have long mean - free paths that are not easily included in hydrodynamic model calculations .
the neutrals must be treated with kinetic equations or monte carlo simulations that must be coupled somehow to the hydrodynamic calculations for the plasma .
* theoretical models of stellar astrospheres may not include all physical processes * : : for example , most models and simulations assume that stellar mass loss is constant or quasi - steady state and , therefore , do not include transient events like coronal mass ejections ( cmes ) that are seen on the sun but not yet detected on stars .
@xcite argued that cmes in active stars could dominate the mass - loss rate based on an extrapolation of solar data , but this process must be tested in realistic models and against stellar cme measurements when they become available .
the magnetic field morphology , in particular the height at which closed field lines open , plays an important role in mass loss .
realistic models should compute the magnetic field structure self - consistently with the mass outflow . *
accurate boundary conditions * : : whether the lism gas flowing into the heliosphere forms a bow shock or a bow wave requires accurate values for the interstellar gas flow speed and magnetic field strength . even for the sun ,
for which we have the best data , these quantities are not yet known accurately .
in particular , the magnetic field strength outside of the heliopause has not yet been measured by the voyager spacecraft or reliably estimated . for nearby stars ,
the flow speed of the star with respect to the interstellar medium gas is probably uncertain by less than 3 km s@xmath9 , but the interstellar magnetic field is not well known . thus the presence or absence of a bow shock is not known in many cases . a critical question is whether the interstellar gas flowing into an astrosphere is partially ionized like the lic or fully ionized . in the latter case
, there will be an astrosphere but no hydrogen wall and thus no possibility for measuring the mass - loss rate .
the presence of lyman-@xmath0 absorption on the short wavelength side of the interstellar absorption indicates the presence of a hydrogen wall and that the star is surrounded by partially ionized interstellar gas , but the inferred value of @xmath5 depends on the neutral hydrogen number density of the inflowing gas which has been assumed to be the same as in the lic given the absence of other measurements .
the nondetection of astrospheric lyman-@xmath0 absorption can be explained either by assuming that the star is embedded in fully - ionized interstellar gas or that the star has an extremely low mass - loss rate .
the latter is unlikely but can not be ruled out at this time . *
active stars appear to have low mass - loss rates * : : we have suggested that the low mass - loss rates for the two rapidly rotating stars with @xmath23 ergs @xmath8 s@xmath9 result from very strong magnetic fields with complex geometries that severely restrict the number of open field lines in the corona .
a new generation of theoretical models is needed to test and develop this scenario .
there are models for very rapidly rotating stars in which magnetocentrifugal acceleration is important @xcite and for t tauri stars for which accretion and disk winds are important ( e.g. , * ? ? ?
what is lacking are models for stars like @xmath19 ori and @xmath17 boo a , which are active stars without extreme rotation or t tauri - like phenomena .
numerical simulations of mhd wave propagation in open flux tubes embedded in realistic atmospheres such as those of @xcite , @xcite , and @xcite provide insight into the essential physics and need to be developed further . observations of more stars , especially those with @xmath12 near and above @xmath24 erg @xmath8 s@xmath9 , are also needed to determine whether the decrease in @xmath5 for @xmath25 erg @xmath8 s@xmath9 is sharp or gradual , or indeed typical . also ,
reobservations of stellar lyman-@xmath0 lines 510 years after the original observations would test whether astrospheres and mass - loss rates change during the time scale for stellar winds to propagate well out into their astrospheres .
the study of astrospheres is intimately connected with the measurements of stellar - wind mass fluxes for late - type stars . at present
the lyman-@xmath0 astrospheres technique is the only successful technique for infering these mass - loss rates .
new theoretical models for stellar astrospheres and stellar mass loss are needed to explain in physical terms the results provided by this technique .
jl acknowledges support by the space telescope science institute ( stsci ) to the university of colorado for hst observing program go-11687 .
bw acknowledges support by stsci for observing program go-12596 .
stsci is operated by the association of universities for research in astronomy , inc . under nasa contract nas 5 - 26555 .
we thank the referees for their very useful comments . emission line ( bottom )
is illustrated in this schematic sketch of the journey of lyman-@xmath0 photons from the star to an observer inside the heliosphere @xcite .
astrospheric absorption is blue - shifted relative to the line of sight component of the interstellar flow velocity because the stellar hydrogen - wall gas is decelerated ( in the stellar reference frame ) relative to the interstellar medium flow .
lyman-@xmath0 photons then traverse the lism and are absorbed by interstellar neutral hydrogen at the projected flow velocity of the lism .
the hydrogen wall gas in the heliosphere produces redshifted absorption as a result of its deceleration relative to the interstellar gas .
the effect of these three absorption components ( illustrated in the middle plots ) is to produce extra absorption on the blue side of the lism absorption indicating the presence of a hydrogen wall surrounding the star , and absorption on the red side of the lism feature produced by the solar hydrogen wall.,width=566 ] lines of six stars with hydrodynamic models assuming wind speeds of 400 km s@xmath9 and parameters of the lism @xcite .
the effect of increasing stellar - wind mass flux is to increase the amount of blue - shifted absorption by the stellar hydrogen wall ( in units of the solar mass loss rate @xmath26 solar mass per year ) .
the absorption line centered near -80 km s@xmath9 is deuterium lyman-@xmath0 absorption in the lism.,width=566 ] erg @xmath8 s@xmath9 , there is a power law relation comparing mass flux rate to x - ray flux .
this simple relation does not hold for stars with larger @xmath12 including the young star @xmath19 uma , @xmath17 boo a , and the m dwarf flare stars proxima centauri and ev lac @xcite.,width=453 ] | charge - exchange reactions between outflowing stellar wind protons and interstellar neutral hydrogen atoms entering a stellar astrosphere produce a region of piled - up - decelerated neutral hydrogen called the hydrogen wall .
absorption by this gas , which is observed in stellar lyman-@xmath0 emission lines , provides the only viable technique at this time for measuring the mass - loss rates of f
m dwarf stars .
we describe this technique , present an alternative way for understanding the relation of mass - loss rate with x - ray emission , and identify several critical issues . |
while most interpretations of the 750 gev diphoton excess recently reported by atlas and cms @xcite assume gluon production ( for a sample of such studies , see for example @xcite ) , a more minimal scenario via photoproduction ( as proposed in @xcite ) is also plausible .
the photoproduction cross section for a 750 gev resonance is able to account for the entire excess provided the resonance has sizable @xmath7 branching ratio to two photons . in this paper
, we simulate the two production scenarios , provide the full cross sections taking into account both elastic and inelastic photoproduction , and present kinematical distributions useful for discriminating between the two scenarios for the production of the hypothetical resonance .
complementary to this paper , ref .
@xcite has studied quark - antiquark annihilation production of the resonance .
the production process of a 750 gev diphoton resonance has observable features in the kinematical distribution of events in the signal region .
in particular , the color flow of photon fusion ( @xmath8f ) or @xmath9 fusion processes would suppress the production of central hadronic activity resulting in fewer central jets compared to a gluon fusion ( @xmath10f ) signal while also enhancing the number of central pseudorapidity gaps , central regions of the detector absent of hadronic activity .
the kinematic properties of the events in the region of the diphoton excess are reported to have no significant difference compared to events above and below the excess diphoton invariant mass region .
we find that this is consistent at 3.2 fb@xmath0 of data with a resonance produced dominantly through either @xmath11f or @xmath10f , assuming the background makes approximately half of the detected events as suggested by the number of diphoton events predicted and observed in the excess region by atlas @xcite .
however , the additional jet multiplicity expected from a @xmath10f signal compared to the expected jet multiplicity from the dominant irreducible diphoton background events would already suggest an approximately 1.5-@xmath12 excess in the total number of jets in the signal region @xmath13 gev for the 38 events measured by atlas if the resonance were produced though @xmath10f .
furthermore , the measurement of the central rapidity gap rate , expected from the signal @xmath8f and background events but exponentially suppressed in @xmath10f events , could discriminate between a vbf or @xmath10f signal but requires special care in order to ensure central tracks originating from pileup do not bury the signature . due to differences in the shape of parton distribution functions and the kinematics between @xmath8f and @xmath10f events
, the resonance tends to be produced more forward in @xmath11f production .
we note that this affects the rapidity distribution of the photons , preferring rapidities corresponding to the detector end caps more often than gluon fusion produced events , although this effect alone is not sufficiently large to fully account for the large end cap excess in the cms diphoton search @xcite .
we find that photons from a decaying heavy scalar resonance are more often central than photons from background events and tend to be detected in the detector barrel rather than the end caps .
if the hypothetical resonance has a large width and the large excess in the 700 gev bin of the cms search is to be considered a signal , then this result is in tension with the interpretation of the large fraction of cms diphoton events with one photon detected in the barrel and one detected in the end cap ( ebee category ) as a scalar resonance signal .
the situation for a wide resonance is improved for a spin-2 resonance for which one expects more events in the ebee category than for a scalar signal .
elastic photoproduction events result in forward and backward protons which can be detected by forward detectors installed by atlas and cms @xcite .
elastic production is suppressed with respect to inelastic .
however , the detection of two intact protons in the final state , with @xmath14 matched to @xmath15 , can be used to remove background .
it was estimated in @xcite that approximately 20 fb@xmath0 is needed for a 5-@xmath12 discovery in this channel . in this paper
we use this luminosity as a benchmark to characterize which features of the production mechanism may be apparent in the kinematic properties of excess events at or before 20 fb@xmath0 of data .
. ] following @xcite
, we will consider a model with an additional scalar particle @xmath16 with mass @xmath17 gev whose only sizable coupling to sm particles is to photons via the operator @xmath18 with @xmath19 gev introduced to have dimensionless couplings , resulting in a partial width to photons @xmath20 of @xmath21 and total width @xmath22 to match the observed event rate , varied between 3 - 6 fb , assuming photon fusion dominates .
the central value ( red ) corresponds to a 4.5 fb production cross section . ] in this paper , we include the contributions from inelastic - inelastic , elastic - inelastic , and elastic - elastic processes ( see fig .
[ fig : photoproduction ] ) . in the narrow width approximation ,
the total photoproduction cross section at @xmath23 tev is @xmath24 determined at leading order ( lo ) from madgraph 5 @xcite with the parton distribution function set nn23lo1 @xcite . for elastic collisions ,
the equivalent photon approximation is made with the improved weizsacker - williams formula @xcite in order to account for the electromagnetic form factor of the proton .
inelastic collisions dominate the production followed by partially elastic and elastic collisions in the ratios 63:33:4 , respectively . here , we see that the rate of @xmath25-@xmath26 fb ( corresponding to the excess number of events observed by atlas ) can be accommodated by a photoproduced resonance with total width of @xmath27 gev , motivated by the best - fit width of the atlas excess , with branching ratio to two photons of approximately 2% .
if we allow the total width @xmath22 to vary , the relationship between @xmath28 and @xmath22 is fixed by matching the observed event rate of the excess and is shown in fig .
[ fig : brvsgamma ] .
we assume @xmath29 gev for the remainder of this paper , although the conclusions still apply for a narrower resonance .
the total production cross section at @xmath30 tev is @xmath31 the ratio @xmath32 determined from madgraph is approximately 2 and does not provide an explanation for the absence of a signal in run i diphoton searches .
however , the ratio for elastic and partially elastic production depends strongly on the finite size effects of the proton , or equivalently the maximum fraction of the proton momentum transferred to an emitted photon , and can be larger than 2 depending on the correct value @xcite .
and gluon coupling @xmath33 to match the observed event rate , varied between 3 - 6 fb , assuming gluon fusion dominates .
the central value ( red ) corresponds to a 4.5 fb production cross section . ] for comparison to the gluon fusion scenario , the effective operator responsible for production is @xmath34 the cross section at @xmath23 tev in the narrow width approximation is given by @xmath35 where the @xmath36factor can be approximated by matching to known nnlo results for a heavy higgs - like scalar @xcite and is taken to be @xmath37 .
the ratio @xmath32 , determined from madgraph , for @xmath38 is 4.5 . matching the diphoton cross section to the observed excess event rate determines the relationship between @xmath28 and @xmath33 as shown in fig .
[ fig : brvscgg ] .
the production mechanism of a 750 gev resonance can be probed by measuring the hadronic observables in the diphoton signal events .
the most prominent features in the @xmath8f events compared to @xmath10f are the suppression of central jets with @xmath39 and the appearance of central pseudorapidity gaps .
these effects are absent for a resonance produced via @xmath10f because the two incoming protons are color - connected , and the color flux tube breaking fills up the central region with soft hadrons from the fragmentation .
however , for a @xmath40-channel exchange of a color - singlet , as in photoproduction , the remnants from each initial proton remain color singlets and only form color connections with partons originating within the same proton , typically resulting in very forward jets ( or intact protons ) with a pseudorapidity gap in the central region .
the two forward jets expected in @xmath9 fusion , from the parton recoil after emitting the heavy gauge boson , are typically not detected in @xmath11f production , although two forward jets would be a signature of a resonance produced via @xmath9 fusion .
these observables have been well - studied for vbf higgs production @xcite . for the higgs ,
@xmath10f dominates the cross section , making the hadronic observables of the vbf events difficult to measure . for a 750 gev photoproduced diphoton resonance ,
however , the event color flow effects of the vbf production should be observable since the small background rate in the diphoton channel does not bury the signal .
one experimental challenge is then to remove pileup which contaminate the events with central hadronic activity . on average , there are between 10 - 20 inelastic collisions per bunch crossing at 2015 run ii luminosities .
atlas and cms have powerful tracking capabilities making possible the matching of jets to the hard interaction and identification of pileup jets @xcite . to remove the jets resulting from pileup ,
only the jets matching the primary vertex of the hard interaction should be considered .
cms has shown the efficiency to match jets with @xmath41 gev to the hard interaction to be @xmath42 in @xcite and to be flat over the range @xmath43 , corresponding to the rapidity acceptance of the tracker .
pileup tracks must also be subtracted from the event in order to observe central rapidity gaps . only tracks matched to the diphoton primary vertex , perhaps by a longitudinal impact parameter cut , should be considered in determining whether an event contains a pseudorapidity gap .
this pileup subtraction procedure may be non - standard , and the presence of central pileup tracks not matched to the primary vertex could bury this signal if special care is not taken . in this section , we present the results of the event generation and fast detector simulation for the two production scenarios of a 750 gev scalar or tensor diphoton resonance and the dominant irreducible diphoton background at @xmath23 tev .
the effective operators in eq . and
eq . have been implemented in model ufo files created with feynrules @xcite .
we also simulate the production of a spin-2 resonance @xmath44 coupled to the stress energy tensor @xmath45 : @xmath46 where @xmath47 and the same definition for @xmath48 but with @xmath49 replaced by @xmath50 .
simulation of the hard process is performed at lo with madgraph 5 @xcite , followed by parton showering with pythia 8 @xcite .
we use delphes 3 @xcite for the fast detector simulation with the default atlas detector geometry and efficiencies .
the irreducible , prompt @xmath8 background is simulated at next - to - leading order ( nlo ) using madgraph@nlo followed by herwig 6 @xcite . at tree level , the only contribution is @xmath51 annihilation . including nlo qcd corrections , this background is the dominant source accounting for more than @xmath52 ( @xmath53 ) of the background near the excess invariant mass region in the ebeb ( ebee ) selection category for the cms diphoton search @xcite .
the reducible background consists of @xmath54 and @xmath55 events in which jet fragments are misidentified as photons .
we expect similar background composition for the atlas diphoton search .
we define an accepted diphoton event for both signal and background as an event containing : * two reconstructed photons , * leading photon @xmath56 gev and subleading @xmath41 gev , * 725 gev @xmath57 775 gev , and * each photon satisfies isolation requirements within a @xmath58 cone .
this invariant mass window is conservative , and the signal to background ratio could be improved with a tighter cut on invariant mass around the resonance . in the accepted diphoton events , tracker and calorimeter information
is used to cluster jets with fastjet @xcite according to the anti-@xmath59 algorithm ( size parameter @xmath60 and minimum jet @xmath61 gev ) .
f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only.,title="fig : " ] f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only.,title="fig : " ] distribution per accepted diphoton event for @xmath10f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only.,title="fig : " ] distribution per accepted diphoton event for @xmath10f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only.,title="fig : " ] f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only . , title="fig : " ] f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only . ,
title="fig : " ] for @xmath10f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only.,title="fig : " ] for @xmath10f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : sample of accepted @xmath10f signal ( black ) and @xmath8f signal ( red ) both combined with 50% @xmath8 background contamination with statistics corresponding to 20 fb@xmath0 .
uncertainties are statistical only.,title="fig : " ] f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : pseudorapidity of photons in accepted diphoton events , normalized by the number of accepted diphoton events , @xmath62 , for @xmath10f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
uncertainties are statistical only.,title="fig : " ] f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
right : pseudorapidity of photons in accepted diphoton events , normalized by the number of accepted diphoton events , @xmath62 , for @xmath10f ( black ) , @xmath8f ( red ) , and irreducible @xmath8 background ( blue ) .
uncertainties are statistical only.,title="fig : " ] the conclusions drawn from hadronic observables concerning the resonance production mechanism for the scalar signal also apply to the tensor case , so we only present hadronic observables of the scalar signal . in figs .
[ fig : njets ] and [ fig : ht ] ( left ) , we present the jet multiplicity and scalar sum of transverse energy , @xmath63 , of accepted diphoton events for @xmath10f and @xmath11f scalar prodution and irreducible background @xmath64 .
these observables demonstrate the additional hadronic activity , also visible in the charged particle multiplicity , in @xmath10f events compared to @xmath8f or background events .
the @xmath10f events prefer 2.0 jets per event , while the @xmath8f and irreducible @xmath11 events have an average of 0.9 and 1.3 jets per event . however , a @xmath10f scenario would not be distinguishable with the existing data due to background contamination .
we expect a 5-@xmath12 statistically significant excess in the total number of jets for the case of a @xmath38 produced resonance at 10 fb@xmath0 of data .
the peak in the number of events for both signals and background at @xmath65 gev is due to the energy of the diphoton . for a @xmath10f produced resonance ,
the @xmath63 distribution of signal events would contain excess events in the @xmath66 gev region , which can be attributed to the extra jets in @xmath10f events . also in figs .
[ fig : njets ] and [ fig : ht ] ( right ) , we present the same distributions for a for a @xmath67-signal , @xmath67-@xmath8 sample of events with statistics corresponding to 20 fb@xmath0 .
the difference in jet multiplicity for @xmath10f vs. @xmath11f production is clearly discernible at 20 fb@xmath0 . in additional to the difference in overall amount of hadronic activity in the two production scenarios ,
there is also different angular dependence due to the event color flow and kinematics of the lab frame . in fig .
[ fig : jeteta ] ( left ) , we show the suppression of central jets for @xmath68 in @xmath11f events .
this effect is somewhat washed out due to the peak in background @xmath8 events at central jet rapidities .
the most striking feature comparing @xmath8f to @xmath10f events is the appearance of central pseudorapidity gaps for the former .
we define a central rapidity gap as the maximum symmetric region of pseudorapidity around @xmath69 for which there are no tracks with @xmath70 .
we choose to only consider tracks in the determination of this observable in order to guarantee the pileup contamination can be removed by tracker information .
the atlas and cms trackers have acceptance out to @xmath71 , giving a maximum observable ( track ) gap size of @xmath72 = 5 . the fraction of signal events containing such a rapidity gap for @xmath73 gev is shown in fig . [ fig : gaps ] ( left ) .
the choice of @xmath74 was made in order to optimize the difference in number of gaps between @xmath10f and @xmath8f signals and to ensure the tracks are high quality in order to be matched to the diphoton vertex , but we note that @xmath75 $ ] gev is useful for discriminating between @xmath10f and @xmath8f production .
the number of events in the @xmath76 bin is large from overflow of events with larger gap sizes ( which can not be measured due to the acceptance of the tracker ) .
large rapidity gaps are exponentially suppressed for @xmath38 events , so a @xmath10f signal component would decay exponentially with @xmath72 compared to ( approximately ) linearly fall of the background with @xmath72 .
the fraction of accepted diphoton events containing a @xmath77 gap are 1.6% , 11.7% , and 19.1% for @xmath10f , @xmath8f , and @xmath8 background events , respectively .
this suggests an enhancement to the rapidity gap rate for the case of @xmath8f production and a suppression compared to background for @xmath10f .
this difference in the predicted number of rapidity gaps in the two scenarios is not statistically significant with the existing data . in fig .
[ fig : gaps ] ( right ) , we show a sample of central rapidity gaps corresponding to 20 fb@xmath0 of data .
one caveat with measuring rapidity gaps is that it relies on modeling of nonperturbative effects and can not be precisely predicted perturbatively . , for @xmath11f production of a spin-0 ( red ) or spin-2 ( purple ) resonance and irreducible @xmath8 background ( blue ) .
right : pseudorapidity of photons in accepted diphoton events for @xmath10f production of a spin-0 ( black ) or spin-2 ( green ) resonance and irreducible @xmath8 background ( blue ) .
uncertainties are statistical only.,title="fig : " ] , for @xmath11f production of a spin-0 ( red ) or spin-2 ( purple ) resonance and irreducible @xmath8 background ( blue ) .
right : pseudorapidity of photons in accepted diphoton events for @xmath10f production of a spin-0 ( black ) or spin-2 ( green ) resonance and irreducible @xmath8 background ( blue ) .
uncertainties are statistical only.,title="fig : " ] the two production modes considered also have different parton distribution functions and kinematics affecting the hard process in the events .
we show these effects in the diphoton and photon pseudorapidity distributions in fig .
[ fig : eta ] .
the diphoton ( resonance ) tends to be more forward produced in @xmath8f events , peaking at @xmath78 . in @xmath10f signal events , @xmath79 peaks at @xmath80 , while the dip in diphoton production at @xmath81 for @xmath10f events is not as pronounced .
the rapidity distribution of the individual photons in fig .
[ fig : eta ] ( right ) is flat for the background but peaked at the center of the detector for both signals suggesting that a scalar diphoton resonance prefers to decay to photons in the detector barrel .
this can be understood as follows . in events where a heavy scalar resonance is produced ,
the decay photons are isotropic in the parton center of mass frame . in the lab frame , the boost factor of the resonance is typically small resulting in an almost isotropic @xmath82 distribution ( with small peaks towards @xmath83 from the small boost factor of the resonance along the beam axis ) for the decay photons .
the shape of the @xmath84 distribution is dominated by the jacobian factor from changing variables from @xmath82 to @xmath84 , which has the same qualitative shape as the photon pseudorapidity distributions for the resonance signals .
the @xmath11 background events , on the other hand , are highly boosted with a @xmath82 distribution that is strongly peaked at @xmath85 . when combined with the jacobian factor ,
the @xmath84 distribution of the @xmath8 events turns out to be flat .
this is not the case for a spin-2 resonance produced via @xmath10f or @xmath11f , which does not decay isotropically to photons in the parton center of mass frame @xcite resulting in more forward photons than a scalar signal . in fig .
[ fig : costheta ] , we present the @xmath84 distributions for @xmath10f and @xmath11f production of a spin-0 and spin-2 resonance , confirming that a tensor resonance gives more forward photons than the scalar signal . .expected fraction of diphoton events accepted by the cms diphoton search @xcite with both photons detected in the barrel ( ebeb category ) or with one photon in the barrel and the other in the end cap ( ebee category ) for @xmath10f and @xmath11f production of either a spin-0 or spin-2 resonance along with irreducible @xmath11 background events . [ cols="<,^,^,^,^,^ " , ] cms reported an unexpectedly large number of the excess events in the ebee selection category , in which one of photons from the diphoton is detected in the end cap ( @xmath86 ) and one in the barrel ( @xmath87 ) , rather than both photons detected in the barrel ( ebeb category ) @xcite . whether these events should be considered as signal events depends on the width and mass of the resonance .
we apply the cuts from the cms diphoton analysis event selection to obtain the expected number of events in the ebee and ebeb categories for the signal and background in table [ tab : tab1 ] .
the signal diphoton events ( for both a @xmath10f or @xmath11f produced scalar resonance ) are most likely to be accepted in the ebeb selection category with @xmath88 of accepted signal events expected in this category .
this is a direct consequence of the @xmath84 distributions in fig .
[ fig : eta ] ( right ) for the @xmath10f and @xmath11f signals . for the spin-2 signal , however , more events are accepted in the ebee category compared to the scalar signal for both @xmath10f and @xmath11f production .
the irreducible background diphotons are more likely to show up in ebee category than the photons from a 750 gev scalar resonance .
we note that the expected number of background events reported by cms in the ebee category is larger than the number expected in the ebeb category , which is not the case for the irreducible @xmath11 background alone .
we understand this effect to be due to be the additional presence of @xmath55 background events , which we do not simulate , that account for @xmath89 of the ebee category expected background but a significantly smaller fraction for the ebeb category .
we have simulated the two production mechanism of a hypothetical 750 gev diphoton resonance at lo and the largest source of irreducible diphoton background at nlo .
the main differences between the two production modes are the jet multiplicities and jet rapidities of the accepted diphoton events in addition to the intact protons from elastic photoproduction , all which may be observed with high statistical significance around or before 20 fb@xmath0 of data .
more subtle features of the signal events include an excess number of events with @xmath90 gev for @xmath38 production and an enhancement to the central pseudorapidity gap rate for @xmath11f production .
moreover , we find that scalar resonance signals strongly prefer to decay to photons in the barrel rather than the endcap , hinting that , for a wide resonance interpretation , the large fraction of excess events accepted in the ebee diphoton selection category by the run-2 cms diphoton search may not be explained by a spin-0 resonance .
if the excess persists with the same features after the collection of more data , a spin-2 resonance may better describe the number of photons detected in the detector end cap .
we would like thank jordan tucker , nathan mirman , and wee hao ng for useful discussions .
c.c . and s.l are supported in part by the nsf grant phy-1316222 .
is supported in part by the doe under grant de - fg02 - 85er40237 .
is supported in part by the doe under grant de - sc-000999 . the atlas collaboration , atlas - conf-2015 - 081 .
cms collaboration , cms - pas - exo-15 - 004 .
k. harigaya and y. nomura , http://arxiv.org/pdf/1512.04850[hep-ph/1512.04850 ] .
s. knapen , t. melia , m. papucci and k. zurek , http://arxiv.org/pdf/1512.04928[hep-ph/1512.04928 ] .
r. franceschini _ et al . _ , http://arxiv.org/pdf/1512.04933[hep-ph/1512.04933 ] .
s. d. mcdermott , p. meade and h. ramani , http://arxiv.org/pdf/1512.05326[hep-ph/1512.05326 ] .
a. falkowski , o. slone and t. volansky , http://arxiv.org/pdf/1512.05777[hep-ph/1512.05777 ] .
p. agrawal , j. fan , b. heidenreich , m. reece and m. strassler , http://arxiv.org/pdf/1512.05775[hep-ph/1512.05775 ] .
b. bellazzini , r. franceschini , f. sala and j. serra , http://arxiv.org/pdf/1512.05330[hep-ph/1512.05330 ] .
d. curtin and c. b. verhaaren , http://arxiv.org/pdf/1512.05753[hep-ph/1512.05753 ] . c. csaki , j. hubisz and j. terning , http://arxiv.org/pdf/1512.05776[hep-ph/1512.05776 ] . s. fichet , g. von gersdorff and c. royon , http://arxiv.org/pdf/1512.05751[hep-ph/1512.05751 ] .
w. altmannshofer , j. galloway , s. gori , a. l. kagan , a. martin and j. zupan , http://arxiv.org/pdf/1512.07616[hep-ph/1512.07616 ] .
j. gao , h. zhang and h. x. zhu , http://arxiv.org/pdf/1512.08478[hep-ph/1512.08478 ] .
m albrow , m arneodo , v avati , j baechler , n cartiglia , m deile , m gallinaro , j hollar , m lo vetere , k oesterberg , n turini , j varela , d wright , and collaboration cms - totem . .
technical report cern - lhcc-2014 - 021 .
totem - tdr-003 .
cms - tdr-13 , cern , geneva , sep 2014 .
j. alwall _ et al .
_ , jhep * 1407 * , 079 ( 2014 ) doi:10.1007/jhep07(2014)079 http://arxiv.org/pdf/1405.0301[hep-ph/1405.0301 ] .
r. d. ball _ et al . _
[ nnpdf collaboration ] , nucl .
b * 877 * , 290 ( 2013 ) doi:10.1016/j.nuclphysb.2013.10.010 http://arxiv.org/pdf/1308.0598[hep-ph/1308.0598 ] .
s. frixione , m. l. mangano , p. nason and g. ridolfi , phys .
b * 319 * , 339 ( 1993 ) doi:10.1016/0370 - 2693(93)90823-z http://arxiv.org/pdf/hep-ph/9310350[hep-ph/9310350 ] .
y. l. dokshitzer , v. a. khoze and t. sjostrand , phys .
b * 274 * , 116 ( 1992 ) .
doi:10.1016/0370 - 2693(92)90312-r t. l. lungov and c. o. escobar , phys .
d * 53 * , 4857 ( 1996 ) doi:10.1103/physrevd.53.4857 http://arxiv.org/pdf/hep-ph/9510209[hep-ph/9510209 ] .
a. de roeck , v. a. khoze , a. d. martin , r. orava and m. g. ryskin , eur .
j. c * 25 * , 391 ( 2002 ) doi:10.1007/s10052 - 002 - 1032 - 9 http://arxiv.org/pdf/hep-ph/0207042[hep-ph/0207042 ] .
cms collaboration , cms - pas - jme-14 - 001 .
the atlas collaboration , atlas - conf-2014 - 018 .
a. alloul , n. d. christensen , c. degrande , c. duhr and b. fuks , comput .
commun . * 185 * , 2250 ( 2014 ) doi:10.1016/j.cpc.2014.04.012 http://arxiv.org/pdf/1310.1921[hep-ph/1310.1921 ] .
t. sjostrand , s. mrenna and p. z. skands , comput .
commun .
* 178 * , 852 ( 2008 ) doi:10.1016/j.cpc.2008.01.036 http://arxiv.org/pdf/0710.3820[hep-ph/0710.3820 ] .
j. de favereau _ et al .
_ [ delphes 3 collaboration ] , jhep * 1402 * , 057 ( 2014 ) doi:10.1007/jhep02(2014)057 http://arxiv.org/pdf/1307.6346[hep-ex/1307.6346 ] .
g. corcella , i. g. knowles , g. marchesini , s. moretti , k. odagiri , p. richardson , m. h. seymour and b. r. webber , jhep * 0101 * , 010 ( 2001 ) doi:10.1088/1126 - 6708/2001/01/010 http://arxiv.org/pdf/hep-ph/0011363[hep-ph/0011363 ] .
m. cacciari , g. p. salam and g. soyez , eur .
j. c * 72 * , 1896 ( 2012 ) doi:10.1140/epjc / s10052 - 012 - 1896 - 2 http://arxiv.org/pdf/1111.6097[hep-ph/1111.6097 ] .
d. boer , w. j. den dunnen , c. pisano and m. schlegel , phys .
lett . *
111 * , no . 3 , 032002 ( 2013 ) doi:10.1103/physrevlett.111.032002 http://arxiv.org/pdf/1304.2654[hep-ph/1304.2654 ] .
the atlas collaboration , atlas - conf-2013 - 029 . c. han , h. m. lee , m. park and v. sanz , http://arxiv.org/pdf/1512.06376[hep-ph/1512.06376 ] . | the production mechanism of a 750 gev diphoton resonance , either via gluon or photon fusion , can be probed by studying kinematic observables in the diphoton events .
we perform a detector study of the two production modes of a hypothetical scalar or tensor diphoton resonance in order to characterize the features of the two scenarios .
the nature of the resonance production can be determined from the jet multiplicity , the jet and diphoton rapidities , the rate of central pseudorapidity gaps , or the possible detection of forward protons from elastic photoproduction for events in the signal region .
kinematic distributions for both signals and expected irreducible diphoton background events are provided for comparison along with a study of observables useful for distinguishing the two scenarios at an integrated luminosity of 20 fb@xmath0 .
we find that decay photons from a 750 gev scalar resonance have a preference for acceptance in the central detector barrel , while background events are more likely to give accepted photons in the detector end caps .
this disfavors the interpretation of the large number of excess events found by the the run-2 cms diphoton search with one photon detected in the end cap as a wide spin-0 resonance signal .
however , one expects more end cap photons in the case of spin-2 resonance .
gluon vs. photon production of a 750 gev diphoton resonance * csaba cski@xmath1 , jay hubisz@xmath2 , salvator lombardo@xmath1 , john terning@xmath3 + * @xmath4department of physics , lepp , cornell university , ithaca , ny 14853 , usa + @xmath5department of physics , syracuse university , syracuse , ny 13244 , usa + @xmath6department of physics , university of california , davis , ca 95616 [email protected] , [email protected] , [email protected] , [email protected] |
recent experiments with bose - einstein condensates ( bec ) driven by a far off - resonant laser beam have demonstrated collective superradiant rayleigh @xcite and raman scattering @xcite , sharing strong analogies with the superradiant emission from excited two - level atoms @xcite . in these experiments an elongated bec scatters the pump photons into the end - fire modes along the major dimensions of the condensate , acquiring a momentum multiple of the two - photon recoil momentum @xmath0 , where @xmath1 and @xmath2 and @xmath3 are the wave vectors of the pump and the scattered field .
theoretical works have shown that the superradiant rayleigh scattering relies on the quantum collective atomic recoil ( qcarl ) gain mechanism , in which the fast escape of the emitted radiation from the active medium leads to the superradiant emission @xcite .
the quantum regime of carl @xcite occurs when the two - photon recoil frequency @xmath4 is larger than the gain bandwidth , such that the recoil frequency shifts the atoms out of resonance inhibiting further scattering processes . as a consequence in the qcarl each atom coherently
scatters a single pump photon , changing momentum by @xmath5 .
the process in which the atoms make a transition between two momentum states ( @xmath6 and @xmath7 ) has strong analogies with that of two - level atoms prepared in the excited state and decaying to the lower state by spontaneous and stimulated emission .
however , the incoherent spontaneous emission dominating the two - level atomic decay is absent in the momentum transition , where spontaneous emission is associated to momentum diffusion due to the scattering force , which can be made very small if the laser is sufficiently detuned from the atomic resonance .
the absence of doppler broadening and the long decoherence time of a bec allows to observe superradiance and coherent spontaneous emission much more easily than from electronic transitions in excited atoms , in which the decay is dominated by the incoherent spontaneous emission .
another example of cooperative phenomena from excited two - level atoms is subradiance , i.e. the cooperative inhibition of spontaneous emission by a destructive interatomic interference .
this phenomenon , whose existence has been proposed by dicke ( 1954 ) in the same article predicting superradiance @xcite , has received less consideration than the more popular superradiance , also due to the difficulty of its experimental observation . in fact , the only experimental evidence has been done on 1985 by pavolini _
_ @xcite . among different schemes of multi - level systems
in which subradiance was predicted , crubellier _ et al .
_ , in a series of theoretical papers @xcite , proposed a three - level degenerate cascade configuration in which cooperative spontaneous emission is expected to exhibit new and striking subradiance effects . in this paper we show that subradiance in a three - level degenerate cascade can be realized in a bec inserted in a ring cavity and lightened by two laser fields with frequency difference twice the two - photon recoil frequency , as illustrated by fig.[fig1 ] . the frequency of the scattered photon is determined by energy and momentum conservation .
the process consists in two steps . in the first step
the atoms initially at rest scatter the laser photons of frequency @xmath8 into the cavity mode of frequency @xmath9 , changing momentum from @xmath10 to @xmath11 . in the second step
the atom scatters the laser photon of frequency @xmath12 changing their momentum from @xmath13 to @xmath14 .
since the change of the kinetic energy of the atom is @xmath15 , by energy conservation the frequency of the scattered photon is @xmath16 which coincides with the frequency generated in the first step when @xmath17 . in this way a three - momentum - level degenerate cascade is realized in which the atoms , initially with momentum @xmath18 , change momentum to the intermediate value @xmath0 and then to the final value @xmath19 , emitting two degenerate photons of frequency @xmath9 . in general the process , as described in ref.@xcite , will continue with an other scattering of the photon of frequency @xmath12 changing the atomic momentum from @xmath19 to @xmath20 with the emission of a photon of frequency @xmath21 and so on .
however , if the cavity linewidth is much narrower than the frequency difference , @xmath22 , these other frequencies will be damped out .
then , the oscillation of only the frequency @xmath9 in the cavity will restrict the momentum cascade to the three momentum states , @xmath10 , @xmath0 and @xmath19 .
a basic feature of this system is that the transition rates are proportional to the pump intensities , so that they can be varied with continuity .
this makes the subradiance observation much easier than with a three - level cascade between electronic energy levels , where the transition rates are fixed by the branching ratios .
the quantum collective atomic recoil laser ( qcarl ) with a two - frequency pump is described by the following equations for the order parameter @xmath23 of the matter field and the cavity mode field amplitude @xmath24 @xcite : @xmath25\psi\label{es}\\
\noindent\frac{da}{dt}&= & gn\alpha(t ) \int dz|\psi|^2e^{i(qz-\delta t)}-\kappa a , \label{ef}\end{aligned}\ ] ] where @xmath26 is the coordinate along the cavity axis and @xmath27 .
these equations have been derived performing the adiabatic elimination of the atomic internal degrees of freedom @xcite but replacing the pump field with @xmath28 . in eqs.([es ] ) and ( [ ef ] ) @xmath29 is the dimensionless electric field amplitude of the scattered radiation beam with frequency @xmath30 , @xmath31 is the coupling constant , @xmath32 is the rabi frequency of the pump laser incident with an angle @xmath33 with respect to the @xmath26 axis ( @xmath34 if counterpropagating ) , with electric field @xmath35 and
frequency @xmath8 detuned from the atomic resonance frequency @xmath36 by @xmath37 .
the pump laser has a sideband with frequency @xmath12 , with @xmath17 [ where @xmath4 and @xmath38 , and electric field @xmath39 with @xmath40 .
the other parameters are : @xmath41 , the electric dipole moment of the atom along the polarization direction @xmath42 of the laser , @xmath43 , the cavity mode volume , @xmath44 the total number of atoms in the condensate , @xmath45 , and @xmath46 , the cavity linewidth .
the emitted frequency @xmath30 is within the cavity frequency linewidth , whereas the pump field is external to the cavity so that its frequencies are not dependent on the cavity ones .
the order parameter @xmath47 of the matter field is normalized such that @xmath48 .
if the condensate is much longer than the radiation wavelength and approximately homogeneous , then periodic boundary conditions can be applied on the atomic sample and the order parameter can be written as @xmath49 , where @xmath50 $ ] are the momentum eigenstates with eigenvalues @xmath51 . using this expansion , eqs.([es ] ) and
( [ ef ] ) become : @xmath52\label{cm } \\
\frac{da}{dt } & = & g n \alpha(t)\sum_{n}c_{m}c^*_{m+1}-\kappa a\label{af},\end{aligned}\ ] ] where @xmath53 .
as as been discussed elsewhere @xcite , if the gain rate is smaller than the recoil frequency the atoms recoil only along the positive direction of @xmath54 , absorbing a photon from the laser and emitting it into the cavity mode .
backward recoil , in which an atom absorbs a photon from the cavity mode and emits it into the laser mode , is inhibited by energy conservation . in this way
, the laser photon of frequency @xmath8 induces a momentum transition from @xmath55 to @xmath56 , emitting in the cavity a photon with frequency @xmath9 ; the laser photon of frequency @xmath12 [ with @xmath17 ] induces a momentum transition from @xmath56 and @xmath58 , emitting an other photon of the same frequency @xmath30 .
if the cavity linewidth @xmath46 is smaller less than @xmath59 , only the photons with frequency @xmath30 will survive in the cavity . since further scattering would generate photons with frequencies @xmath60 , with @xmath61 , which can not oscillate in the cavity , then the hilbert space of the atoms
is spanned by only the first three recoil momentum levels , @xmath62 , and eqs.([cm ] ) and ( [ af ] ) reduce to : @xmath63\label{c1 } \\
\frac{d c_2}{d t}&=&2i(\delta-2\omega_r)c_2 + g\alpha(t ) a^*c_{1}\label{c2 } \\
\frac{da}{dt } & = & g n \alpha(t)(c_{0}c^*_{1}+c_{1}c^*_{2})-\kappa a\label{a3}.\end{aligned}\ ] ] eqs.([c0])-([a3 ] ) contain fast oscillating terms .
they can be eliminated introducing the slowly varying variable @xmath64 and approximating eqs.([c0])-([a3 ] ) neglecting the fast oscillating terms proportional to @xmath65 . in this way
eqs.([c0])-([a3 ] ) reduce to : @xmath66 eqs.([c0:2])-([a3:2 ] ) describe the three - level degenerate cascade of the atoms driven by two laser fields at frequencies @xmath8 and @xmath67 , respectively , and interacting with the self - generated cavity mode at the frequency @xmath9 .
notice that the second transition rate , from @xmath56 to @xmath58 , is proportional to the two pump amplitude ratio , @xmath68 .
asymptotically , in a time much longer than @xmath69 , the photons leak the cavity and the total polarization in eq.([a3:2 ] ) vanishes : @xmath70 on resonance ( @xmath71 ) and with the atoms initially at rest ( @xmath72 ) , the variables @xmath73 , @xmath74 , @xmath75 and @xmath76 are real and eqs.([c0:2])-([a3:2 ] ) keep invariant the following quantity : @xmath77 from it we see that the atoms can not populate completely the final state @xmath58 ( with @xmath78 and @xmath79 ) unless @xmath80 .
hence , when @xmath81 the atoms remain in the intermediate levels @xmath55 and @xmath56 in a subradiant state . condition ( [ pol ] ) , together with the constraints ( [ inv ] ) and the normalization @xmath82 , determine univocally the steady - state solution reached asymptotically by the atoms .
it is easy to show that for @xmath83 , @xmath84^{1/2}\ ] ] whereas for @xmath85 , @xmath86 fig.[fig2 ] shows the steady - state populations @xmath87 for @xmath88 plotted vs. @xmath68 .
we observe that increasing @xmath68 from zero the population of the intermediate state , @xmath89 , decreases and the population of the final state , @xmath90 , increases .
they are equal for @xmath91 , with @xmath92 and @xmath93 .
the population of the initial state @xmath94 increases too and reaches a local maximum for @xmath95 with @xmath96 , @xmath97 and @xmath98 .
then , for @xmath85 the intermediate state @xmath56 is empty ( @xmath97 ) and the population of the initial state @xmath94 decreases to zero for @xmath99 ; then for @xmath100 it increases until it equals the population of the final state @xmath90 when @xmath101 . for @xmath102
the population of the initial state , @xmath94 , is larger than that of the final state , @xmath90 .
however , this case appears stationary only because the semiclassical model neglects spontaneous emission .
the quantum treatment , reported in the next section , shows that the stationary subradiant state may exist only for @xmath103 .
( dashed red line ) , @xmath89 ( dotted gray line ) and @xmath90 ( continuous blue line ) as a function of @xmath68 , as given by eqs .
( [ sc:1 ] ) and ( [ sc:2]).,width=302 ] 0.2 cm in order to illustrate how the system evolves toward the subradiance state , fig.[fig3 ] shows the time evolution of the field @xmath104
( fig.[fig3]a ) , and the three populations , ( fig.[fig3]b ) [ @xmath94 ( dashed blue line ) , @xmath89 ( red continuous line ) and @xmath90 ( dashed - dotted black line ) ] , obtained solving the complete equations ( [ cm ] ) and ( [ af ] ) for @xmath105 , @xmath106 , @xmath71 and @xmath91 , with initial condition @xmath72 , @xmath107 and @xmath108 . the final populations are @xmath109 and @xmath110 , according to eq .
( [ sc:1 ] ) . , ( a ) , and the three populations @xmath94 ( dashed blue line ) , @xmath89 ( red continuous line ) and @xmath90 ( dashed - dotted black line ) vs. @xmath111 , where @xmath112 , for @xmath105 , @xmath106 , @xmath71 and @xmath91.,width=9 ] 0.2 cm in order to test the dependence of the subradiance state on the frequency difference @xmath113 between the two pump fields , figure [ fig4 ] shows the asymptotic coherence @xmath114 between the intermediate and the final states vs. @xmath113 for @xmath115 , @xmath71 , @xmath91 , @xmath116 ( dashed blue line ) , @xmath106 ( dashed - dot red line ) and @xmath117 ( continuous black line ) .
the result shows that subradiance requires a very fine tuning of the pump frequency difference near @xmath59 , within a precision @xmath118 . between the intermediate and the final states vs. @xmath119 for @xmath115 , @xmath71 , @xmath91 , @xmath116 ( dashed blue line ) , @xmath106 ( dashed - dot red line ) and @xmath117 ( continuous black line ) .
] 0.2 cm as a second example , figs .
[ fig5 ] and [ fig6 ] show that same case as in figs.[fig3 ] and [ fig4 ] but with @xmath101 . in this case @xmath120 and @xmath121 .
we note that whereas in the case @xmath91 the resonance linewidth of fig.[fig4 ] decreases when the cavity losses @xmath46 increases , on the contrary in the case @xmath101 the linewidth increases with @xmath46 and it is about a factor @xmath122 larger .
hence the subradiance with @xmath91 is more sensible to the frequency mismatch than that with @xmath101 .
( a ) and the three populations @xmath94 ( dashed blue line ) , @xmath89 ( red continuous line ) and @xmath90 ( dashed - dotted black line ) as a function of @xmath111 , where @xmath112 , for the same parameters of fig.[fig3 ] and @xmath101 . ]
0.2 cm between the initial and the final states vs. @xmath119 for @xmath115 , @xmath71 , @xmath101 , @xmath116 ( dashed blue line ) , @xmath106 ( dashed - dot red line ) and @xmath117 ( continuous black line ) . ]
let now obtain the subradiant state quantum - mechanically , treating the amplitude @xmath123 as bosonic operators @xmath124 with commutation rules
@xmath125=\delta_{m , n}$ ] . then , according to eq.([pol ] ) the subradiant state @xmath126 satisfies : @xmath127 ( we omit the tilde on @xmath128 ) .
it is possible to demonstrate ( see appendix [ app1 ] ) that for a system of @xmath44 atoms ( with @xmath44 even ) there are @xmath129 subradiant states @xmath130 , with @xmath131 , defined as @xmath132 where @xmath133 and @xmath134 is a normalization constant . the index @xmath135 is related to the population difference between the initial and final states , since @xmath136 .
the case @xmath137 corresponds to the state @xmath138 .
the link between the subradiant state @xmath130 and the semiclassical solution ( [ sc:1 ] ) and ( [ sc:2 ] ) is provided by the correspondence between @xmath135 and the population difference @xmath139 in the limit @xmath140 . for @xmath83 , @xmath141 and for @xmath85 ,
@xmath142 $ ] .
as particular cases , @xmath143 for @xmath91 and @xmath144 for @xmath95
. furthermore , @xmath137 for @xmath80 and @xmath145 for @xmath101 .
hence @xmath146 is the maximum value of @xmath68 , giving the following subradiant state : @xmath147 for large @xmath44 , the average value of @xmath148 is @xmath149 , with variance @xmath150 . from the state ( [ state ] ) and the correspondence between @xmath135 and @xmath68 we may evaluate the average populations @xmath151 , with @xmath88 , as a function of @xmath68 . the result is compared in fig.[fig7 ] with the semiclassical solution ( [ sc:1 ] ) and ( [ sc:2 ] ) , for @xmath152 .
we observe that the quantum solution has not a sharp transition at @xmath95 as the classical one , but there a tail which becomes negligible for @xmath153 .
( red circles ) , @xmath89 ( black squares ) and @xmath90 ( blue triangles ) vs. @xmath68 , obtained from eq .
( [ state ] ) with @xmath152 .
the dashed lines show for comparison the semiclassical solution of fig.[fig2 ] .
, width=264 ] 0.2 cm here we show the wigner function of the subradiance state @xmath126 in order to get some more properties of the system .
we start from the definition @xmath154 where @xmath155 and @xmath156 are complex numbers and @xmath157 is the characteristic function defined as @xmath158 where @xmath159 is a displacement operator for the @xmath160-th mode . a straightforward calculation , reported in the appendix [ app2 ] , yields @xmath161 and @xmath162 where @xmath163 and @xmath164 is the laguerre polynomial .
notice that the wigner function depends only on the modulus of @xmath155 and not from its phase .
as expected , in general it is negative due to the presence of the laguerre polynomials . by integrating over the other two mode variables , from eq.([wigsr ] )
we obtain the single - mode wigner functions : @xmath165 in order to investigate the characteristics of the subradiance state , let s consider some specific example .
an interesting case is when @xmath91 and @xmath143 , for which the semiclassical theory yields @xmath166 .
fig.[ex1](a ) shows the probability @xmath167 vs. @xmath148 for @xmath168 and @xmath169 .
the probability is maximum for @xmath170 , the average value is @xmath171 and the standard deviation is @xmath172 .
the single - mode wigner functions @xmath173 are shown in fig.[ex1](b - d ) : @xmath174 has a maximum at @xmath175 and @xmath176 and @xmath177 have a maximum at @xmath178 , in agreement with the population values predicted by the semiclassical solution .
however , @xmath176 differs considerably from @xmath177 with a strong oscillation near @xmath179 , probably due to the tail of the distribution at small @xmath148 observed in fig.[ex1](a ) .
the single - mode wigner functions present a pronounced maximum around which they are positive , plus an oscillating quantum background . as a second example we consider the case @xmath101 , for which the semiclassical theory yields @xmath120 and @xmath180 . in the quantum model it corresponds to the maximally anti - symmetric state ( [ state : n/2 ] ) with @xmath145 .
fig.[ex2](a ) shows the probability @xmath167 vs. @xmath148 for @xmath168 and @xmath181 .
the probability is maximum for @xmath182 and decreases rapidly to zero for larger @xmath148 , with @xmath183 and @xmath184 .
the single - mode wigner functions @xmath185 are shown in fig.[ex2](b - d ) . @xmath174 and @xmath177 are equal and very similar to the wigner function of the number state @xmath186 , @xmath187 @xcite . furthermore , @xmath176 is equal to the vacuum wigner function @xmath188 . in this case
@xmath129 pairs of atoms with momentum @xmath10 and @xmath189 are produced . ,
@xmath169 and @xmath91 : ( a ) probability @xmath167 vs. @xmath148 ; ( b ) @xmath174 vs. @xmath190 ; ( c ) @xmath176 vs. @xmath191 ; ( d ) @xmath177 vs. @xmath192.,width=321 ] 0.2 cm , @xmath181 and @xmath101 : ( a ) probability @xmath167 vs. @xmath148 ; ( b ) @xmath174 vs. @xmath190 ; ( c ) @xmath176 vs. @xmath191 ; ( d ) @xmath177 vs. @xmath192.,width=321 ] 0.2 cm the three - mode wigner function ( [ wigsr ] ) , after integrating one mode variable , yields the following two - mode wigner functions : @xmath193 figures [ figw01 ] , [ figw02 ] and [ figw12 ] ( color online ) show the two - mode wigner functions @xmath194 as a function of @xmath195 and @xmath196 , for @xmath197 for the case @xmath168 and @xmath169 , corresponding to @xmath91 ( see fig.[ex1 ] ) : the red color corresponds to a negative value of the functions .
we observe several zones of negativity indicating non classical correlations between the two modes .
in particular , @xmath198 and @xmath199 show strong correlations of the modes @xmath10 and @xmath200 with the mode @xmath201 , which has strong oscillations , whereas @xmath202 is larger near the vacuum value @xmath203 .
hence , we can say that qualitatively the modes @xmath10 and @xmath200 behave quasi - classically whereas the mode @xmath201 has feature similar to a number state . figures [ figw02a ] , shows @xmath202 vs. @xmath190 and @xmath192 for the case @xmath168 and @xmath181 , corresponding to @xmath101 ( see fig.[ex2 ] ) .
the two - mode wigner function looks the product of two single - mode number wigner functions shown in fig.[ex2](b ) and
( d ) , with a bi - dimensional regular mesh of positive and negative zones . for @xmath168 , @xmath169 . ]
0.2 cm for @xmath168 , @xmath169 . ] 0.2 cm for @xmath168 , @xmath169 . ] 0.2 cm for @xmath168 , @xmath181 . ]
0.2 cm we calculate now the equal - time intensity correlation and cross - correlation functions , defined respectively as : @xmath204 with @xmath88 , @xmath205 and @xmath206 . for a classical field
there is an upper limit to the second - order equal - time cross correlation function given by the cauchy - schwartz inequality @xmath207^{1/2}.\ ] ] quantum - mechanical fields , however , can violate this inequality and are instead constrained by @xmath208^{1/2 } \left[g_j^{(2)}(0)+\frac{1}{\langle \hat n_j\rangle}\right]^{1/2}\label{ineq}\ ] ] which reduces to the classical results in the limit of large occupation numbers .
we obtain the following expressions for the subradiant state : @xmath209 where @xmath210 and @xmath211 the cross - correlation functions are @xmath212 @xmath213 from ( [ g0])-([g12 ] ) it follows that @xmath214 showing that @xmath215 are consistent with the quantum inequality ( [ ineq ] ) .
fig.[fig8 ] shows the intensity correlation functions @xmath216 for @xmath88 , for the subradiance state ( [ state ] ) .
we obtain that @xmath217 ( continuous red line ) is less than unity for all the values of @xmath68 and @xmath218 ( dotted blue line ) is less than unity for @xmath219 .
so anti - bunching occurs for the momentum states @xmath55 and @xmath58 , but not for the state @xmath56 .
fig.[fig9]-[fig11 ] show the eventual violation of the cauchy - schwartz inequality ( [ ineq : cl ] ) for @xmath220 ( continuous line ) .
the dashed line is the classical upper limit @xmath221^{1/2}$ ] and the dotted line is the quantum upper limit @xmath222^{1/2}$ ] .
we found that @xmath223 is always consistent with the classical inequality , whereas @xmath224 and @xmath225 violate the cauchy - schwartz inequality ( [ ineq : cl ] ) for all the @xmath68 and for @xmath226 , respectively , showing the existence of quantum correlations between the modes @xmath55 and @xmath58 and between the modes @xmath56 and @xmath58 .
for @xmath85 , @xmath224 is close to the upper limit of the quantum inequality ( [ ineq ] ) . :
@xmath218 ( dotted blue line ) , @xmath227 ( dashed black line ) and @xmath217 ( continuous red line).,width=264 ] 0.2 cm ( continuous red line ) vs. @xmath68 .
the dashed black line indicates the classical upper limit of the inequality ( [ ineq : cl ] ) ; the dotted blue line indicates the quantum upper limit of the inequality ( [ ineq ] ) .
, width=283 ] 0.2 cm but for @xmath228.,width=302 ] 0.2 cm but for @xmath229.,width=302 ] 0.2 cm
we have investigated a possible way to observe subradiance in a bose - einstein condensate in a high - finesse ring cavity , scattering photons from a two - frequency pump laser into a single - frequency cavity mode via the quantum collective atomic recoil lasing ( qcarl ) mechanism .
subradiance occurs in a degenerate cascade between three motional levels separated by the two - photon momentum recoil @xmath0 , where @xmath230 is the momentum transfer between pump and cavity mode .
the observation of subradiance in momentum transitions of cold atomic samples presents several advantages and differences with respect to the electronic transitions of excited two - level atoms .
first , the momentum transitions are not affected by the spontaneous emission if the pump laser is sufficiently detuned from the atomic resonance .
second , the atomic condensates have a long life and a very long coherence time , allowing the preparation and the further manipulation of the subradiant state .
third , the subradiance is realized among collective motional states containing a large number of atoms .
subradiance as well superradiance do not need that the dimension of the sample is smaller than the radiation wavelength , as in superradiance by excited atoms .
for these reasons , subradiance between motional states of ultracold atoms may be important for the study of the decoherence - free subspaces sought in quantum information @xcite .
other recent proposals of realizing subradiance in matter wave require a very fine control of single atoms in optical cavities @xcite , which can be very problematic experimentally . on the other hand ,
the experimental activity on superradiant rayleigh scattering and carl with bose - einstein condensates @xcite has achieved important progresses and a subradiance experiment with bec in a ring cavity could be feasible with the present day techniques . at ultracold temperature and with a coupling constant @xmath231 much less than the recoil frequency @xmath232 should be possible , using two laser fields with frequency difference @xmath59 , to restrict the momentum transition to only the first two recoil momentum states .
recent experiments on superradiant scattering from a bec pumped by a two - frequency laser beam @xcite have shown that the momentum transitions are enhanced by the presence of the second pump detuned by @xmath59 . however , only inserting the bec in a high - finesse ring cavity it will possible to limit the transition sequence to only two .
then , varying the relative intensity of the two pump laser beams should be possible to probe the transition from superradiance to subradiance . as an example of possible parameters , subradiance could be observed in a ring cavity similar to that realized in tbingen @xcite ( with length @xmath233 mm , beam waist @xmath234 and finesse @xmath235 , about 5 times the presently achieved value ) with @xmath236 and @xmath237 .
this last value can be obtained using a @xmath238rb condensate ( with @xmath239khz ) with @xmath240 atoms at a temperature of some tens of nk , driven by two laser beams with @xmath241mw , @xmath242thz and a frequency difference precision of @xmath243khz .
subradiance will be reached in about @xmath244s , less than the decoherence time of the bec .
in order to demonstrate eq.([state ] ) let s consider a state of the form @xmath245 where the second sum is over all the pairs @xmath246 such that @xmath247 . when substituted in ( [ sr ] )
, it yields @xmath248 after having redefined the indexes @xmath249 and @xmath250 in the sums , it becomes @xmath251 the terms in the curl bracket of eq .
( [ s:2 ] ) vanish when the following conditions are met : 1 . in the second sum on @xmath249 in the curl
bracket the term with @xmath252 is missing : since in the first sum @xmath253 when @xmath252 , it yields @xmath254 .
2 . in the first sum on @xmath249 in the curl bracket the term with @xmath255 is missing : then the second sum on @xmath249 with @xmath255 , @xmath256 and @xmath257 yields @xmath258 3 .
the remaining sum in ( [ s:2 ] ) yields : @xmath259 with @xmath260 . in the second sum of eq.([s:2 ] ) the term @xmath261 is missing , so in the first sum the term with @xmath261 and @xmath262 vanishes and yields @xmath263 with @xmath264 . together with the result of item 1 , it yields @xmath265 whereas the remaining terms of eq.([s:5 ] ) yield @xmath266 with @xmath260 and @xmath267 . for @xmath268 eq.([s:7 ] ) yields @xmath269 since from eq.([s:4 ] ) @xmath270 , then @xmath271 continuing with all the even values of @xmath249 , it is easy to show that @xmath272 hence , the only terms different from zero are those with @xmath273 and @xmath274 , yielding : @xmath275 where @xmath276 .
eq.([s:11 ] ) provides a recurrence relation for the index @xmath277 with a given @xmath278 .
in fact the difference between the first and second index of @xmath279 in both the left and right terms of eq.([s:11 ] ) is @xmath280 .
so , introducing the new index @xmath281 , eq.([s:11 ] ) can be written , for @xmath282 , as : @xmath283 by iteration of eq.([s:12 ] ) we obtain @xmath284 where @xmath285 and @xmath286 . since @xmath287 , finally we obtain : @xmath288 where @xmath289^{1/2}f(p , n - p)$ ] .
we demonstrate eq.([chsr ] ) . writing the displacement operator as a product of operators , @xmath290 and using the formula @xmath291 where @xmath292 is the laguerre polynomial of order @xmath148 . using ( [ dis ] ) in ( [ chfun ] ) with the subradiant state ( [ state ] )
we obtain @xmath293 in order to evaluate the wigner function ( [ wignerdef ] ) we must calculate an integral of the form : @xmath294 introducing polar coordinates @xmath295 and @xmath296 it transforms into @xmath297 where @xmath298 is the bessel function of zero order . using the formula @xcite : @xmath299 we obtain @xmath300 so , from the definition of wigner function ( [ wignerdef ] ) and using eqs . ( [ char ] ) and ( [ ii ] )
we obtain @xmath301 which coincides with eq.([wigsr ] ) . using the formula @xcite @xmath302 we have @xmath303 from ( [ ii ] ) , ( [ wigner ] ) and ( [ norma ] )
we obtain the expressions ( [ wig1])-([wig3 ] ) and ( [ wig01])-([wig12 ] ) of the one - mode and two - mode reduced wigner functions , respectively .
99 s. inouye , a.p .
chikkatur , d.m .
stamper - kurn , j. stenger , d.e .
pritchard and w. ketterle , science * 285 * , 571 ( 1999 ) .
s. inouye , t. pfau , s. gupta , a.p .
chikkatur , a. grlitz , d.e . pritchard and w. ketterle ,
nature * 402 * , 641 ( 1999 ) .
m. kozuma , y. suzuki , y. torii , t. sugiura , t. kugam , e.w .
hagley , l. deng , science * 286 * , 2309 ( 1999 ) .
l. fallani , c. fort , n. piovella , m.m . cola , f.s .
cataliotti , m. inguscio , and r. bonifacio , phys .
a * 71 * , 033612 ( 2005 ) .
d. schneble et al.,phys .
a * 69 * , 041601(r ) ( 2004 ) .
y. yoshikawa et al . , rev .
a * 69 * , 041603(r ) ( 2004 ) .
m. gross and s. haroche , phys . rep . *
93 * , 301 ( 1982 ) .
n. piovella , m. gatelli and r. bonifacio , optics comm .
* 194 * , 167 ( 2001 ) .
robb , n. piovella , and r. bonifacio , j. opt .
b : quantum semiclass . opt .
* 7 * 93 ( 2005 ) .
n. piovella , m. gatelli , l. martinucci , r. bonifacio , b.w.j .
mcneil , and g.r.m .
robb , laser physics , * 12 * , 188 ( 2002 ) .
r. bonifacio and l. de salvo souza , nucl .
and meth . in phys .
a * 341 * , 360 ( 1994 ) .
r. bonifacio , l. de salvo , l.m .
narducci and e.j .
dangelo , phys .
a * 50 * , 1716 ( 1994 ) .
dicke , phys . rev . * 93 * , 99 ( 1954 ) .
d. pavolini , a. crubellier , p. pillet , l. cabaret , and s. liberman , phys .
lett . * 54 * , 1917 ( 1985 ) .
a. crubellier , s. liberman , d. pavolini , and p. pillet , j. phys .
b : at . mol .
* 18 * , 3811 ( 1985 ) .
a. crubellier , and d. pavolini , j. phys .
b : at . mol .
* 19 * , 2109 ( 1986 ) .
a. crubellier , j. phys .
b : at . mol .
20 * , 971 ( 1987 ) .
a. crubellier , and d. pavolini , j. phys .
b : at . mol .
* 20 * , 1451 ( 1987 ) .
m.m . cola , l. volpe , and n. piovella , phys .
a * 79 * , 013613 ( 2009 ) .
n. piovella , m. m. cola and r. bonifacio , phys .
a * 67 * , 013817 ( 2003 ) .
d.f.walls , and g.j .
milburn , _ quantum optics _ , ( springer , berlin , 1994 ) p. 64 .
lidar , i.l .
chuang , and k.b .
whaley , phys .
* 81 * , 2594 ( 1998 ) .
p. fldi , m.g .
benedict , and a. czirjk , phys .
a * 65 * , 021802(r ) ( 2002 ) .
s. slama , s. bux , g. krenz , c. zimmermann , and ph.w .
courteille , phys .
lett . * 98 * , 053603 ( 2007 ) .
n. bar - gill , e.e .
rowen , and n. davidson , phys .
a * 76 * , 043603 ( 2007 ) .
f. yang , x. zhou , j. li , y. chen , l. xia , and x. chen , phys .
a * 78 * , 043611 ( 2008 ) .
s. slama , g. krenz , s. bux , c. zimmermann , ph.w .
courteille , phys .
a * 75 * , 063620 ( 2007 ) .
i.s . gradshteyn and i.m .
ryzhuk , _ table of intagrals , series and products _ , ( academic press , san diego , 2000 ) , p. 803 - 805 | subradiance , i.e. the cooperative inhibition of spontaneous emission by destructive interatomic interference , can be realized in a cold atomic sample confined in a ring cavity and lightened by a two - frequency laser .
the atoms , scattering the photons of the two laser fields into the cavity - mode , recoil and change their momentum . under proper conditions
the atomic initial momentum state and the first two momentum recoil states form a three - level degenerate cascade .
a stationary subradiant state is obtained after that the scattered photons have left the cavity , leaving the atoms in a coherent superposition of the three collective momentum states . after a semiclassical description of the process
, we calculate the quantum subradiant state and its wigner function .
anti - bunching and quantum correlations between the three atomic modes of the subradiant state are demonstrated . |
star formation in external galaxies are intimately linked to questions concerning galaxy formation and evolution .
of particular interest are galaxies with star bursts . because formation is taken place in interstellar clouds , the uv and optical stellar light is re - processed into infrared photons and it is necessary to observe at infrared / millimeter wavelengths and interpret the observations by dust and radiative transfer models .
a two component model of the dust in the interstellar medium ( siebenmorgen & krgel , 1992 ) has found some success in the last years . this model computes the emission per unit mass of dust heated by radiation of known intensity and spectral distribution .
the dust consists of a mixture of large grains , very small grains and organic molecules .
we assume here that the molecules are polycyclic aromatic hydrocarbons ( pah ) , though other band carriers have been suggested ( e.g. sakata et al .
1984 , duley 1988 , papoular et al . 1989 ) . to consider somehow the shape of the bands a lorentzian profile with some damping constant
was suggested by siebenmorgen et al .
such a simple oscillator model can account for observations taken at moderate spectral resolution .
one typical example demonstrating the importance of pah in the mid infrared as well as of large grains but in the far ir is shown in fig.[ring_sed.ps ] .
the galactic nuclei are deeply hidden by dust so that it necessary to carry out radiative transfer calculations in some form of approximation .
we applied the numerical code by krgel & siebenmorgen ( 1994 ) which considers in a consistent computation also a distribution of stars in the galactic nuclei .
because some of the massive stars are so luminous each immediate environment of an ob star presents a hot spot where the dust temperature has a local peak . at those hot spots the heating of the dust
is dominated by the stars while outside the hot spots the dust is heated by the interstellar radiation field ( isrf ) . in fig.[starburst.ps ] we present some of our iso data together with submillimeter observations .
the data are fit quite well with the dust and radiative transfer model described above .
in particular note the strong silicate absorption best seen in the log - log diagrams .
the silicate absorption feature is quite difficult to be recognized from the pht - s data alone . for all galaxies the model parameters give some tentative description of the nucleus and its structure ( siebenmorgen et al .
however , one should also not forget about its limitations : computations are done in a broken three dimensional but still spherical grid .
the real configuration of the nucleus is evidently not spherical so that the structure of the model becomes oversimplified on scales smaller than 400 pc .
m. isophot ( @xmath10 ) and iras data are color corrected .
the isophot error bars reflect the absolute photometric calibration uncertainty .
photometry at 1.3 mm by chini et al .
we show the best fit ( solid curve ) is separated into a 29.9k cold ( dotted ) and a 12.8k very cold dust component ( dashed ) .
figure adopted from siebenmorgen ( 1999b ) .
+ , scaledwidth=100.0% ] ratio can be used as a good discriminator of the activity type of a galaxy .
+ , scaledwidth=100.0% ]
the far ir emission of the starburst galaxies presented in fig.[starburst.ps ] are well fit by a single modified black body at about 32k .
siebenmorgen et al .
( 1999b ) have studied the far ir / mm emission of another sample of 8 galaxies extracted from the markarian catalog ( gardner 1995 ) .
all of them show a similar fir / mm spectral energy distribution .
the spectral energy distribution of the inactive galaxy ugc2936 is shown fig.[u2936_sed.ps ] .
our best fit which is constrained by iso , iras and iram data reveals the presence of second component .
we need to employ a 29.9k cold and a 12.8k very cold dust component .
our multi - filter far ir photometric study of 7 spirals shows that at least two dust components are needed to account for the spectra . beside a cold dust component of about 32k an additional component of very cold dust of about 12k
is present .
a simple average over all markarian and another average over all inactive galaxies of our iso sample give us generic spectra for both types of galaxies .
they are shown in fig.[kalterstaub_sum.ps ] .
siebenmorgen et al .
( 1999b ) therefore confirm the original claim by krgel et al .
( 1998 ) of the existence of such an additional component of very cold dust in spiral galaxies .
determining the gas mass from 1.3 mm dust continuum maps that cover the optical extent of the inactive spirals we find a ratio of total infrared luminosity to gas mass @xmath11 l@xmath5/ m@xmath5 , which is a factor 30 lower than found for the markarian galaxies .
this demonstrates that the @xmath12 ratio can be used as a discriminator of the activity type of a galaxy .
* question : _
p. barthel _ * + could it be that cold dust is present in the mkn galaxies but warm dust is just swamping it ?
+ * answer : * + both galaxy samples are at about similar distance scale and have been observed with the same instrument sensitivities .
the ( very ) cold dust can be best studied in the submillimeter .
there is no or only marginal evidence for source extension at 1.3 mm for the mkn sample whereas the spirals have similar extension as the optical diameter .
therefore the observations do not give evidence of a very cold dust component for the active sample but for the spirals .
+ chini r. , krgel e. , lemke r. , ward
thompson d. , 1995 , a&a 295 , 317 duley w.w .
, williams d.a .
, 1988 , mnras 231 , 969 gardner j.p . , 1995 ,
apjs 98 , 441 , 1995 geballe t.r . , 1991 ,
mnras 251 , 24 krgel e. , siebenmorgen r. , 1994 , a&a 288 , 929 krgel e. , siebenmorgen r. , zota v. , chini r. , 1998 , a&a 331 , l9 sakata a. , wada s. , tanabe t. , onaka t. , 1984 , apj 287 , l5 siebenmorgen r. , krgel e. , 1992 , a&a 259 , 614 siebenmorgen r. , natta a. , krgel e. , prusti t. , 1998 , a&a 339 , 134 siebenmorgen r. , krgel e. , and v. zota , 1999a , a&a 351 , 140 siebenmorgen r. , krgel e. , and r. chini , 1999b , a&a 351 , 495 | the mid infrared spectra of the starbursts show the 9.7@xmath0 silicate absorption feature and strong mid infrared emission bands centered at 6.2 , 7.7 , and 11.3 @xmath1 m .
illustrative models of the _ active _ galaxies are presented : as the starbursts are most likely confined to the central region of the galaxy the radiative transfer in the nucleus is solved under the assumption of spherical symmetry .
the distribution of stars and dust are adjusted until the complete infrared spectrum of the galaxies are modeled .
the dust is described as a mixture of large grains , very small grains and pah , which are undergoing temperature fluctuations .
although the galactic nuclei are deeply hidden in the dust its global structure can be estimated by the simple calculations presented .
isophot samples of eleven _ active _ galaxies and seven _ inactive _ spirals are presented .
the far infrared and submillimeter spectrum of the _ active _ galaxies can be described by a single modified black
body at a color temperature of @xmath2k .
this leads to a ratio of infrared luminosity to gas mass , @xmath3 , of @xmath4l@xmath5/m@xmath5 .
in contrast , the spectral energy distributions of _ inactive _ spirals require , apart from warm dust of @xmath6k , an additional very cold component of at most @xmath7k .
this implies a @xmath3 ratio of @xmath8l@xmath5/m@xmath5 for the _ inactive _ spirals , a factor @xmath9 lower than for the _ active _ galaxies .
the detection of such cold dust can be predicted by radiative transfer models .
dust , radiative transfer , active galaxies , spirals |
motivated by our earlier work on the scattering of topological solitons on potential obstructions in a class of models [ 1,2 ] we have decided to look at the generality of these results . to do this
we have decided to investigate the scattering properties of non - topological solitons so in this paper we look at the scattering of q - balls on various potential obstructions .
first we recall that q - balls are non - topological soliton fields , first introduced by s. coleman in [ 3 ] .
they arise in some classes of self - interacting complex scalar fields . unlike topological solitons ,
whose stability is ensured by the presence of a conserved topological charge , q - balls carry a conserved noether charge associated with the global @xmath6 symmetry .
this conserved but non - topological charge stabilises the q - balls .
also , q - balls have a time - dependent rotating phase and the conserved noether charge is related to the velocity of the change of this phase .
these features result in a much more complicated scattering properties of these solitons , leading to such phenomena as , for example , the charge fission during the scattering process . in this paper
we investigate the q - ball dynamics in @xmath0 and @xmath1 dimensions , mainly when the q - balls are scattered on various obstructions in the form of potential barriers or holes .
first we discuss the conditions of the stability of the q - balls .
then we report the results of our numerical simulations which have revealed that q - balls behave like topological solitons only for the values of the parameters of the theory that guarantee this stability throughout the scattering . when this condition is satisfied the dynamics of the scattering on barriers and holes in @xmath7 dimensions is similar to what was seen for topological solitons [ 1,2 ] , _ ie _ very little radiation is given off during the scattering . the scattering on barriers results in transmission or reflection , _ ie _ very much like that of a small particles . in the scattering on holes , like for topological solitons
, we can have transmissions , trappings and reflections . as discussed before [ 1 - 2 , 4 - 5 ] the trappings correspond to the case of solitons which , while in the hole , lose energy through radiation and the reflections correspond to the very coherent interaction of solitons with this radiation .
we find similar properties of the scattering of q - balls except that this time they also generate further small q - balls .
this scattering is discussed in detail in sections 3 and 4 . in section 5
we discuss the scattering of solitons in @xmath1 dimensions .
we find that , again , the scattering often resembles the scattering of small particles but the properties of this scattering depend crucially on the parameters of the theory .
however , there are also small differences which we discuss in detail .
the paper ends with a short section presenting our conclusions .
we consider a complex scalar field @xmath9 in @xmath8 spatial dimensions [ 6 ] whose lagrangian density is given by @xmath10 where the indices @xmath11 and @xmath12 run from 0 to @xmath8 .
the energy momentum tensor takes the form : @xmath13 where @xmath14 is the space - time metric with signature @xmath15 .
this lagrangian has been so chosen that it is invariant under the global @xmath6 transformation and the conserved noether current,@xmath16 , associated with this symmetry , is given by : @xmath17 the corresponding conserved noether charge,@xmath18 is then @xmath19 the total energy and momentum are given , respectively , by @xmath20 @xmath21 where @xmath22 is the spatial index which runs from 1 to @xmath8 .
the stationary q - ball located at the origin is described by : @xmath23 where @xmath24 is the internal rotation frequency and @xmath25 is the real radial profile function which satisfies the equation of motion which follows from ( 1 ) , namely : @xmath26 where @xmath8 is the number of spatial dimensions . for the consistency of the equations and in order that the q - ball has a finite energy , the profile function @xmath27 , has to satisfy the following boundary conditions @xmath28 and @xmath29 the corresponding field ( 7 ) is then called the stationary q - ball . as the lagrangian ( 1 ) is lorentz invariant the field ( 7 ) can be boosted with any velocity up to the speed of light .
note that the equation for @xmath27 , ( 8) , can be treated as a mechanical analogue describing a point particle moving in an effective potential with friction if @xmath30 is interpreted as the position of the particle and @xmath31 as the time coordinate [ 6 ] .
the effective potential in which this particle moves is then @xmath32 the ` friction ' term @xmath33 becomes important for high @xmath8 or small @xmath31 . first let us choose our potential @xmath34 .
we want it to be such that it gives rise to an exact q - ball solution in @xmath0 dimensions .
this potential should have an absolute minimum when @xmath35 .
then @xmath36 would be the ground state of the theory .
moreover , we require that the symmetry of the theory remains unbroken
. however , looking at ( 11 ) we note that there are further conditions on the parameters of the q - ball ( _ ie _ @xmath24 ) . to see this we observe that the q - ball field exists provided that @xmath37 corresponds to a local maximum of the effective potential _
@xmath38_{f=0}<0 $ ] .
this leads to an upper bound on the frequency @xmath39 .
also , the local minimum of @xmath40 $ ] must be attained at some positive value of @xmath41 and this requirement puts a lower bound on @xmath24 _ ie _ @xmath42 .
@xmath43 < \omega < \,u^{\prime\prime}\left(0\right).\ ] ] thus the q - ball field exists for @xmath24 in this range , _ ie _ @xmath44 .
of course , as ( 11 ) is quadratic in @xmath24 , we can also take negative values of @xmath24 - the corresponding fields are called anti - q - balls .
however , the existence of a q - ball does not guarantee its stability . for a q - ball field configuration to be stable
we have to require that @xmath45 , where @xmath46 is the mass of the scalar field .
the potential we have chosen to use has the form [ 7 ] : @xmath47 with this potential q - ball solutions exist for @xmath48 .
moreover , these q - balls are stable as @xmath49 .
next we insert our obstructions .
following our earlier work [ 1 - 2 ] on the soliton scattering on obstructions we introduce a potential parameter that perturbs the q - ball only in a certain region of space . the q - ball ,
as it moves towards this region , will only experience this perturbation when its profile function @xmath27 in non - zero in this region .
we introduce the perturbation in the form of potential barriers and holes , and hence we modify the potential @xmath50 to become @xmath51 in which @xmath52 . thus @xmath53 is the extra potential parameter which has been inserted into the potential @xmath54 to take into account the effects of obstructions , holes and barriers , and so is non - zero only in a certain region of space . in our case , we have put the obstruction around the origin ( _ ie _ @xmath55 ) and so we have taken @xmath56 the q - ball in our model is really quite sizeable ( _ ie _ the solution of ( 8) for the potential ( 13 ) is nonzero for a sizeable region of @xmath57 ) so the obstruction has been made quite large to make sure that the q - ball size fits into it quite easily ; _ ie _ is significantly larger than the size of the q - ball .
the equation of motion , in the presence of the obstruction , now takes the form @xmath58 using the q - ball ansatz , the field equation for the profile function , @xmath59 now takes the form : @xmath60 this profile equation ( 16 ) can only be solved exactly when @xmath61=constant . then with the help of [ 8 ] we note that @xmath27 is given by @xmath62^{\frac{1}{2}}.\ ] ] as the theory is lorentz invariant this solution can be be boosted or shifted by a spatial translation .
then , the q - ball solution becomes @xmath63^{\frac{1}{2}}\ ] ] where @xmath64 is the position of the q - ball , @xmath65 its velocity and @xmath66 is the usual relativistic factor . however , @xmath67 constant and so this expression does not solve ( 16 ) for all values of the parameters . on the other hand our @xmath68 , ( 18 ) , is strongly localised and so if the q - ball is far away from the obstruction the solution to ( 16 ) is approximately given by ( 18 ) and so ( 18 ) can be used for the initial condition of our numerical simulation of ( 15 ) . note
that when the q - ball is far enough from the obstruction @xmath69 and ( 18 ) becomes @xmath70^{\frac{1}{2}}\ ] ] the associated q - ball noether charge and the energy become @xmath71 @xmath72 where @xmath73^\frac{1}{2}$ ] and @xmath74 .
moreover , @xmath75 , @xmath76 , + @xmath77 we see that both @xmath78 and @xmath79 depend on @xmath24 and so , in follows , we call them @xmath80 and @xmath81 .
thus the q - ball noether charge is @xmath82^\frac{1}{2}\ ] ] and the energy becomes @xmath83.\ ] ] when @xmath84 ( _ ie _ the q - ball is at rest ) the energy becomes its mass and is given by @xmath85 looking at these formulae we observe that as @xmath86 , @xmath87 and @xmath88 and when @xmath89 then @xmath90 and @xmath91 .
1 shows that @xmath92 and @xmath80 are monotonically decreasing functions of @xmath24 .
furthermore , as @xmath24 increases the charge decreases and in the limit as @xmath93 the charge , @xmath94 . in this upper bound limit
the energy - charge ratio approaches the mass of the scalar field , _ ie _ @xmath95 . in the lower bound limit , _ ie _ when @xmath96 , the ratio @xmath97 .
2 shows clearly that the energy - charge ratio is below the mass term of the scalar field when the velocity of the q - ball is small [ 9 ] .
when @xmath98 the energy - charge ratio is always below the stability line and only intersects this line when @xmath99 .
thus , the q - balls are stable for small velocities and so are prevented from decaying into a number of smaller q - balls .
however , when the q - ball acquires higher speeds the absolute stability condition can be violated ; q - balls can still be long - lived and will , eventually , decay into a number of smaller q - balls [ 7 ] .
2 shows that as @xmath24 increases and so @xmath100 and @xmath87 decrease the stability condition becomes more violated for higher velocities . when an obstruction is present in a certain region of space the mass term gets modified by the perturbation factor @xmath3 _ ie _ @xmath101 .
thus , the stability line shown in fig .
2 gets shifted and the amount of the shift depends on the magnitude of the perturbation . in the case of a barrier
the stability line gets shifted upwards and downwards in the case of a hole .
consequently , the q - ball is less stable in the presence of holes .
this is what we have observed in our simulations .
this scattering on holes is very delicate since if we want the q - ball to be stable against fission , as the q - ball reaches the hole , the hole must be relatively shallow .
so , when the potential is lowered by a positive factor ( but less than one ) , _ ie _ @xmath102 , the tail of the q - ball may develop some small q - balls in the hole even when the q - ball has a vanishing velocity . on the other hand ,
the q - ball stability is not affected by the height of the barrier in cases when the velocity is small .
in fact , within some limits , the higher the barrier the more stable is the q - ball scattering as the stability line is shifted further upwards .
however , we have to be careful ; the q - ball may also become unstable if the barrier is too high . and the charge @xmath87 as a function of @xmath24
. , width=377 ] / @xmath78 as a function of @xmath24 for various velocities.,width=453 ] in next two sections we discuss the dynamics of q - balls as they meet an obstruction located around @xmath103 . in order to understand it we had to perform some numerical simulations .
these simulations were performed using the 4th order runge - kutta method of simulating the time evolution .
we used 12001 points with the lattice spacing of @xmath104 .
hence , the lattice extended from -60 to 60 in the @xmath57-direction .
the time step was chosen to be @xmath105 . in our work
we used the absorbing boundary conditions .
when a q - ball , or a topological soliton , are far away from an obstruction , to a good approximation , they can be treated as a point particle following a well defined trajectory . however , when they approach the barrier the differences in their behaviour begin to arise .
these differences stem from the fact that the q - charge of the q - ball is not quantized and that not only q - balls exist only in a certain range of frequencies but also there are some conditions on the q - ball that have to be satisfied for it to be stable .
the work on topological solitons [ 1 - 4 ] has revealed that the speed and the magnitude of the barrier do not affect significantly the general dynamics of the solitons .
topological solitons suffer some distortions during their scattering process but they do recover their original shape after the scattering as the scattering is quite elastic . for the q - balls , however , their dynamics on obstructions ( barriers or holes ) is connected to their stability .
and this stability depends on the velocity , the height of the obstruction and the internal frequency of the q - balls and thus on their charge .
one major difference between the q - balls and topological solitons as they scatter on barriers is their behaviour close to their critical velocities .
critical velocities of the solitons are the lowest velocities above which they can overcome the barrier and get transmitted and below which they are reflected .
all models involving topological solitons that we have studied have shown that the solitons can be treated as point particles and that such a discussion leads to very accurate estimates of their critical velocity .
there are no restrictions on the speed of the solitons or the height of the barrier . in the case of the q - balls the situation is different .
first of all , the critical velocity can not be determined without having to impose some restrictions on their speeds and the height of the barrier .
this is due to the possibility of the q - ball becoming unstable during its interaction with the barrier . as we have seen from fig .
2 the stability condition is satisfied only for low velocities . when a barrier is present , the stability line is raised and more q - ball field configurations become stable .
so to begin with we will consider the scattering on very small barriers .
before we go on and discuss the scattering dynamics of the q - ball on such barriers , we have to determine whether the force between the q - ball and the barrier is repulsive or attractive .
to check this we have placed a q - ball at rest close to the barrier .
3 presents a trajectory of a q - ball with @xmath106 initially placed at rest at @xmath107 , _ ie _ five units of distance from a barrier of height 0.01 .
the plot clearly demonstrates that the q - ball is repelled by the barrier .
this repulsion is due to the tail of the q - ball which interacts with the barrier .
the further away we place q - ball the smaller the tail and hence the weaker the repulsion but the repulsion is always there .
so to study the scattering on the barrier we need to give the q - ball a boost . , @xmath108 , placed at @xmath107 from a barrier of 0.01 height , width=377 ] we have performed several simulations by boosting the q - balls with various small velocities .
all these simulations have shown that , like for the topological solitons , the scattering is very elastic .
there is very little radiation sent out and although the shape of the q - ball changes as it climbs the barrier - afterwards it returns to its original shape . looking at the dependence on the frequency ( _ ie _ @xmath78 ) of the q - ball we have noticed that the q - balls of higher value of @xmath78 get closer to the barrier .
however , when we tried to determine the velocity at which they reach the top of the barrier ( _ ie _ the critical velocity ) this dependence on the frequency disappears .
thus for a barrier of height 0.01 the numerically found value of @xmath109 was @xmath110 .
as the scattering is very elastic we have tried to derive this value in the point particle approximation . in this case
we look at the energy of the q - ball at rest at the top of the barrier and compare it with the energy of a moving q - ball away from the barrier . if the scattering is elastic the equality of these energies gives us a lower bound on the critical velocity .
thus in our case @xmath111 this energy should be very close to the rest mass energy of a q - ball at the top of a barrier given by @xmath112 thus the critical velocity is given by @xmath113 so , let us use the above formula to estimate the critical velocity .
the rest mass energy of the q - ball with @xmath106 , far away from the potential barrier , is @xmath114 while its rest mass at the top of a barrier of height 0.01 is @xmath115 .
thus , @xmath116 , which is in an excellent agreement with the numerical value mentioned before .
this was for the q - ball of @xmath106 .
what about other values ? in table 1 .
we present the calculated critical velocities for several values of @xmath24 . [ cols= " <
, < , < , < , < " , ] table 4 .
these angles have been calculated numerically by using @xmath117 where @xmath118 and @xmath119 are the velocities along the @xmath57- and @xmath120- axes , respectively .
how do these results depend on the height of the barrier or the frequency of the q - ball ? or on the velocity of the q - ball ?
we have performed further studies and have found that when a q - ball is sent at higher speed , the deflection angle is smaller .
this is clearly due to the fact that at higher velocity the q - ball has more power to overcome the barrier and so gets nearer to the centre of the barrier .
what about the dependence on the height of the barrier ?
16 presents the trajectories of q - balls of frequency @xmath121 sent from @xmath122 towards the barriers of height 0.5 and 0.9 .
we note that the deflection angle is smaller for the smaller barrier ; again , in this case the q - ball gets nearer to the centre of the barrier .
sent from @xmath123 with velocity @xmath124 towards barriers of height 0.9 and 0.5.,width=529 ] furthermore , we have also found that the deflection angle increases with the decrease of the q - ball charge ( _ ie _ @xmath24 ) .
this is shown in fig .
17 in which we present plot of trajectories of two q - balls ( with @xmath121 and 1.9 ) sent from @xmath125 with velocity 0.1 towards a barrier of height 0.9 located at @xmath126 .
the deflection angles for the q - balls with @xmath121 and @xmath106 are found to be 31.955 and 40.167 , respectively . and @xmath106 sent from @xmath125 with @xmath124 towards a barrier of height 0.9 located at @xmath127,width=529 ] we have also studied the q - ball scattering on holes . like in the ( @xmath128 )
dimensional case the whole process is more complicated than the scattering on barriers . as in the @xmath129 dimensional case the complications arise from the generation of ` baby ' q - balls in the hole and their repulsion .
also the smaller the velocity the more time the ` parent ' q - ball has for generating these ` baby ' q - balls .
this , coupled with the dependence on the impact parameter , makes the whole process very complicated .
so here we present our preliminary results - leaving more detailed investigations for the future . however , these preliminary results are interesting enough to report them here .
first of all let us note that , like in the ( @xmath130 dimensional case , the overall force between the q - ball and the hole can be repulsive , attractive or a mixture of these two and this depends on the velocity of the q - ball , its charge and the position and the depth of the hole .
this dependence is mainly due to the conditions of the stability of the q - ball .
when the q - balls , during the process of the scattering , remain stable the force between them and the hole is purely attractive and the process resembles the scattering on the barriers except that this time the deflection is towards the hole .
however , the behaviour depends on the initial velocity ; the faster the q - balls move the less they are deflected by the hole .
18 presents the trajectories of 3 q - balls ( of @xmath131 ) sent with velocities @xmath132 and @xmath133 when the hole ( of depth -0.9 ) was located at @xmath134 .
the plots show very clearly that the force between the q - ball and the hole is attractive but the deflection decreases with the increase in the speed of the q - ball . sent with various incoming velocities for a hole of depth -0.9 located at @xmath135.,width=529 ] next we report the results of our study of the dependence on the impact parameter .
19 presents the trajectories of the q - ball ( @xmath136 sent with velocity @xmath137 at various impact parameters , _ ie _ @xmath138 , and @xmath139 . as expected , we see that the deflection increases as we decrease the impact parameter ( when the q - balls are sent with the same velocity ) . for a hole of depth -0.9 located at various impact parameters.,width=529 ] as we decrease the impact parameter @xmath140 further , bringing the hole closer to the @xmath141-axis , the net interaction between the q - ball and the hole ceases being purely attractive and becomes partially repulsive and and partially attractive .
this results in a complicated behaviour which is due to what we have seen in the one - dimensional scattering : the hole is no longer a passive object but rather a charged object ( due to the ` baby q - balls ' that have been generated in the hole ) and that affects the interaction . fig .
20 presents the trajectory of the ` parent ' q - ball ( of @xmath131 ) sent with velocity 0.2 towards a hole located at @xmath142 .
the figure shows very clearly that initially the scattering is repulsive and later becomes attractive .
note the small scale of the @xmath120-axis which shows that the motion in the @xmath143 direction is very small .
we have also checked that if we increase the velocity the repulsive behaviour disappears .
hence the behaviour is very sensitive to the values of various parameters ( velocity or impact parameter ) . ) sent with velocity @xmath144 towards a hole of depth -0.9 located at @xmath142.,width=529 ] furthermore , when a q - ball is sent with a small velocity , it can interact with the hole earlier . at this stage
the force is attractive and the q - ball has not yet had the time to generate many ` baby ' q - balls in the hole .
thus the repulsive force is smaller or even disappears altogether and a purely attractive force is observed .
21 presents the trajectories of 2 q - balls ( with @xmath131 ) sent with velocities @xmath145 @xmath146 towards a hole located at @xmath142 . from the figure one sees very clearly that the repulsive force seen in the case of @xmath146 disappears when
the q - ball velocity is decreased to @xmath147 and that , in this latter case , the overall force is attractive . ) sent with velocities @xmath146 and @xmath147 towards a hole of depth -0.9 located at @xmath142.,width=529 ] the magnitude and the type of the overall force between the q - ball and the hole , when a q - ball is sent with a certain speed and at a certain impact parameter , depends also on the charge of the q - ball ( _ ie _ @xmath24 ) .
22 presents the trajectories of q - balls of different charges ( different @xmath24 ) when each q - ball was sent with velocity 0.1 and towards a hole of -0.9 located at @xmath148 .
we note a very strong dependence on the charge . sometimes the trajectories are deflected towards the hole , for other @xmath24 they are deflected in the other direction .
the amount of the deflection also depends strongly on @xmath24 .
clearly this is a complicated process which requires further investigation .
one factor that certainly is partially responsible for this behaviour is the broadness of the profile functions .
however , this broadness is basically monotonic ( but not at the tail ( see fig . 14 ) ) so it is hard to predict how much charge ( in the form of ` baby ' q - balls ) get generated in the hole and how this affects the overall scattering properties .
all this behaviour requires more detailed further investigation
. sent with @xmath137 , a hole of -0.9 depth located at @xmath148 , width=529 ] finally , we have also checked , numerically , that the forces are independent of the overall helicity , that is the force does not depend on the sign of @xmath24 .
the behaviour is independent of whether we send a q - ball or an anti - q - ball at a hole located at the same position .
the dynamics also does not depend on the sign of the impact parameter ; _ ie _ it is also still the same when a q - ball is sent ( with the same velocity ) towards a hole located at a corresponding negative impact parameter .
as the forces between a q - ball and a hole are attractive in the region of stability of the q - ball all deflections can be cancelled if we send a q - ball symmetrically in - between two holes .
then the only effect is a small attraction as the q - ball approaches the line joining the holes followed by a repulsion .
so , in this case the q - ball behaves again like a point particle .
in this paper we have presented the results of our studies of the scattering of q - balls on potential obstructions ( in the form of potential barriers and holes ) in ( @xmath130 and @xmath149 dimensions . in our work
we have tried to compare our results with similar results for topological solitons . and in many cases the results were similar ; however , some results were different as q - balls have some conditions that they have to satisfy to be stable .
moreover , as their charge is not quantised they can , in the interaction with obstructions , generate further small q - balls ( called ` baby ' q - balls in this paper ) .
we have found that in the barrier case the scattering of q - balls resembled very closely the scattering of topological solitons .
the scattering was very elastic .
for each barrier there was a critical velocity below which the q - ball was reflected and above which it was transmitted .
what was surprising , at first sight , was that the value of this critical velocity did not depend on the charge of the q - ball ( _ ie _ it was independent of @xmath24 ) .
as the scattering was very elastic we could estimate the value of this critical velocity by performing an energy analysis ( _ ie _ by comparing the energy of a boosted q - ball away from the barrier to the energy of q - ball at rest at the top of the barrier ) .
the results were in an excellent agreement with what was seen in the numerical simulations .
this suggests that , indeed , the scattering is very elastic and the approximation of q - balls by point - like particles is quite reliable . for the scattering of q - balls on potential holes the situation was slightly different as the stable q - balls , when they fall into a hole can become unstable
this instability translates into the generation of ` baby ' q - balls which can then interact with their ` parent ' .
we have performed detailed studies of the scattering of q - balls on potential holes . when the q - balls are stable , and remain stable in the hole , they , again , resemble topological solitons . in this case
they can either fall into the hole and get trapped in it or pass through it with a slightly reduced velocity .
the scattering is , again , quite close to being elastic although this time more radiation is produced than in the case of a barrier .
however , when a q - ball approaches the hole , especially when its parameters are close to the stability bounds , it produces small q - balls in the hole . as the basic force between q - balls can be of either sign ( it depends on their relative phase ) the ` baby ' q -
balls then start interacting with each other and with the initial q - ball
. the overall interaction can now be attractive or repulsive and this depends on the parameters of the configuration ( velocity and frequency of the q - ball , the position and the height of the hole etc ) . in ( @xmath150 )
dimensions we have looked at the scattering of a q - ball on a series of potential barriers whose position was varied along the @xmath120 axis .
the q - ball was deflected from the potential barrier .
we have found that the deflection angle depended on the impact parameter , the height of the barrier , the velocity and the charge @xmath151 of the q - ball . .
we have also studied the interactions of stable q - balls with the hole in ( @xmath150 ) dimensions at various impact parameters .
we have found the expected deflection towards the hole although the trajectories of the ` parent ' q - ball were slightly irregular .
this irregularity increased at the lower impact parameters and it depended also on the velocity of the q - ball .
when we studied the dependence of the impact parameter on the charge of the q - ball ( _ ie _ the frequency @xmath24 ) we have found a surprising irregularity as shown in fig .
22 . for some frequencies the deflection was towards the hole , for some others away
and it varied in magnitude in a very unpredictable way .
clearly the overall forces in this case are very finely balanced .
it is clear that to get a good understanding of the scattering on a hole requires further work .
al - alawi , w.j .
_ j. phys .
_ * a 40 * , 11319 - 1131 ( 2007 )
al - alawi , w.j .
zakrzewski , _
_ * a 41 * , 315206 ( 2008 ) .
s. coleman , _ nucl .
* b 262 * , 263 ( 1985 ) .
b. piette and w.j .
zakrzewski , _ j. phys . _
* a 40 * , 329 - 346 ( 2007 ) . b. piette and w.j .
zakrzewski , _ j. phys .
_ * a 40 * , 5995 - 6010 ( 2007 )
tsumagari , e.j .
copeland and p.m. saffin , some stationary properties of a q - ball in arbitrary space dimensions , hep - th/0805.3233v2 ( 2008 ) .
m. axenides , s. komineas , l. perivolaropoulos , m. floratos , _ phys . rev .
_ * d 61 * , 085006 ( 2000 ) .
a. kusenko , phys . lett .
* 405 b * , 108 ( 1997 ) .
bowcock , d. foster and p.m. sutcliffe , q - balls , integrability and duality , hep - th/0809.3895v1 ( 2008 ) . | we discuss various scattering properties of non - topological solitons , q - balls , on potential obstructions in @xmath0 and @xmath1 dimensions .
these obstructions , barriers and holes , are inserted into the potential of the theory via the coupling parameter , @xmath2 @xmath3 , that is effective only in a certain region of space . when @xmath4 the obstruction is a barrier and when @xmath5 the obstruction is a hole .
the dynamics of q - balls on such obstructions in @xmath0 dimensions is shown to be very similar to that of topological solitons provided that the q - balls are stable . in @xmath1 dimensions ,
numerical simulations have shown some differences from the dynamics of topological solitons .
we discuss these differences in some detail . |
revealing the mass accretion process onto supermassive black holes ( smbhs ) is crucial for understanding the nature of agns because it is believed that the mass accretion fuels a smbh , supports its nuclear luminosity , and presumably powers its outflow .
given the simplest approximation of an axisymmetric adiabatic accretion flow onto a smbh , bondi ( 1952 ) derived a black hole accretion power @xmath10 , where @xmath11 is the bondi accretion rate which is defined at the sphere of gravitational influence of the smbh ( i.e. , the bondi radius @xmath12 ) .
the active galaxy m 87 , one of the well - known llagns , possesses a huge black hole mass @xmath13@xmath14 ( macchetto et al .
1997 ; gebhardt et al .
2011 ; walsh et al .
2013 ) with a prominent relativistic jet 10@xmath15 for the black hole mass in this paper . ] .
the estimated jet power is in a range between @xmath16 to @xmath17 erg s@xmath9 ( li et al .
2009 and references therein ) and seems to match fairly well with the overall energetics described by the bondi accretionyr@xmath9 is adopted here . ]
( @xmath18 erg s@xmath9 ) at @xmath19 pc . however , the central core of m 87 is highly under - luminous in the x - ray band and its x - ray luminosity is smaller than @xmath20 by five orders of magnitude ( @xmath21 erg s@xmath9 ; di matteo et al .
the low x - ray luminosity implies the existence of radiatively inefficient accretion flows ( riafs ) and/or a substantial decrease of the mass accretion rate @xmath22 at radii @xmath23 . the theory of riafs , such as the advection - dominated accretion flow ( adaf : ichimaru 1977 ; narayan & yi 1995 ) , convection - dominated accretion flow ( cdaf : narayan 2000 ; quataert & gruzinov 2000 ) , and adiabatic inflow - outflow solution ( adios : blandford & begelman 1999 ) was developed over the last decades .
the resulting mass accretion rates toward an smbh normalized by the bondi accretion rate can be scaled as a function of the spherical radius as @xmath24 with @xmath25 . considering the viscosity parameter @xmath26 ( shakura & sunyaev 1973 ) , solutions of classical adafs suggest a mass accretion rate from @xmath12 to @xmath27 that is comparable to @xmath11 : @xmath28 ( gadaf stands for `` giant '' adaf ; narayan & fabian 2011 ) . on the other hand , cdafs exhibit a substantial decrease of the mass accretion rate toward @xmath29 as @xmath30 ( igumenshchev & narayan 2002 ) .
the adios , which generalizes the adaf model by including the disk wind , can take intermediate values of @xmath31 anywhere between 0 ( adaf ) and 1 ( cdaf ) .
recent numerical simulations favor @xmath320.7 which is consistent with adios ( e.g. , pang et al .
2011 ; yuan et al .
2012 ) .
there has been a growing consensus during the last decade that millimeter ( mm)/submillimeter ( submm ) polarimetry provides useful diagnostics to infer @xmath33 of riafs at @xmath34 ( agol 2000 ; quataert & gruzinov 2000 ) . in particular , through the faraday rotation measure ( rm ; an integral of the product of the thermal electron density and the magnetic field component along the line of sight ) of the linear polarization , @xmath35 toward sgr
a * was examined ( e.g. , bower et al .
2003 ; marrone et al . 2006 ; macquart et al . 2006 ) and an upper limit of @xmath6 was found to be @xmath36@xmath37yr@xmath9 at @xmath38 ( @xmath27 is the schwarzschild radius ) .
we note that the submm wavelengths are more advantageous than other radio bands for determining the rm because the bandwidth depolarization is less significant , and the opacities of the accretion flow and the jet are smaller so that one can probe regions closer to the bh .
however , efforts to determine the @xmath33 of agns ( including m 87 ) have never been undertaken except for sgr a*. for m 87 , all previous efforts on measuring the rm primarily focused on either the jet or radio lobes at centimeter wavelengths .
owen et al . ( 1990 ) found rm values of a few thousands rad / m@xmath5 at lobes ( interpreted as arising from a foreground faraday screen ) and of a few hundreds rad / m@xmath5 in the kiloparsec - scale jet .
more recently , algaba et al .
( 2013 ) found similar rm values with a hint of gradients across the kpc jet . on upstream
parsec scales , junor et al .
( 2001 ) found an average rm of @xmath1@xmath24400 rad / m@xmath5 with regions that show both positive and negative signs .
zavala & taylor ( 2002 ) found similar results and explained the origin of the observed rm with foreground medium not directly associated with the jet . to our knowledge ,
no mm / submm polarization observations for examining rm toward m 87 have been conducted . in this letter , we present the first measurement of rm toward the m 87 nucleus at submm wavelengths .
we also derive the mass accretion rate @xmath6 onto the black hole based on the measured rm and provide a constraint on riaf models . in section 2 , we introduce our sma observation of the rm together with our data reduction . in section 3 ,
we show our main results .
the estimation of @xmath6 , the discussion on the preferred riaf models , and alternative interpretations of rm are presented in section 4 .
m 87 was observed in the 230 ghz band on 2013 january 23 with the submillimeter array ( sma ; ho et al .
the observation was conducted with seven antennas in the extended array configuration .
the weather condition was excellent during the observation with an atmospheric opacity @xmath39 .
the total length of time on m 87 was 8 hours .
the sma receivers operate in double sideband mode with each sideband having a width of 4 ghz .
each sideband was further split into a pair of 2-ghz - wide intermediate frequency ( if ) sub - bands .
we centered those ifs at 230.3 and 232.3 ghz in the upper sideband ( usb ) and at 218.4 and 220.4 ghz in the lower sideband ( lsb ) , respectively .
the sma polarimeter was used to sample all four polarized correlations ( ll , lr , rl , and rr ) by switching polarization between the left - hand and right - hand circularly polarized feeds in period with 16 walsh function patterns . a detailed discussion of the sma polarimetry system is given in marrone ( 2006 ) and marrone & rao ( 2008 ) .
we performed initial flagging and calibrations including the flux , bandpass , and gain calibrations in the mir - idl package developed for the sma .
we conducted polarization calibration , imaging , and data analysis in the miriad package ( wright & sault 1993 ) .
the flux calibration was done with measurements of callisto , and we performed bandpass , amplitude , and phase calibration with frequent observations of the quasar 3c279 , which is [email protected] away from m 87 . to perform polarization calibration , we observed 3c279 and 3c84 over a large range of parallactic angles and solved for quasar polarization and leakage terms .
the average values of the instrumental polarization ( d - terms ) or `` leakage '' were approximately 1@xmath22% in the upper sideband and 4@xmath25% in the lower sideband .
the leakage terms derived from 3c279 and 3c84 were in excellent agreement with each other .
we estimated the accuracy of the leakage terms to be @xmath10.5% .
after deriving the d - terms , we corrected the gain of m 87 data by self - calibration ( phase - only ) techniques in order to increase the dynamic range , and then applied polarization leakages .
we performed imaging and deconvolution with i , q , u , and v imaged individually for each if band in each sideband , leading to four sets of i , q , u , v images .
the final synthesized beams for the continuum images are 1``.2@xmath30''.8 ( corresponding to @xmath1 ( 1.7 @xmath3 1.0 ) @xmath310@xmath4@xmath41 ) .
finally , we used the task impol in miriad along with the q and u maps to derive total linearly polarized intensity and distribution of the electric vector position angles ( evpa ) .
the magnetic field distribution of m 87 were obtained by rotating the evpas by 90@xmath42 .
in figure 1 we show a sample stokes images of m 87 in the 232.3 ghz band .
figure 2 shows the magnetic field distribution .
one can see that the magnetic vector orientations are perpendicular to the jet in knots a and c and parallel in knot b. this is in good agreement with polarization characteristics derived from observations at centimeter wavelengths and optical band ( e.g. , owen et al . 1989 , perlman et al . 1999 ) , indicating the robustness of our polarization observation and calibration . in table 1
, we show the i , q , u fluxes , polarization fractions , and the evpa of the m 87 core for each of the four bands in the observation . in the following
, we will interpret the change of the polarization position angle @xmath0 as a result of _ external _ faraday rotation associated with accretion flow .
the reason why we prefer this interpretation will be explained in section 4.2 .
faraday rotation changes the observed @xmath0 as a function of observing wavelength @xmath43 according to @xmath44 where @xmath45 is the intrinsic polarization angle and rm is the rotation measure .
we determine the rm of the m 87 core by fitting the polarization position angle ( @xmath0 @xmath46 0.5@xmath47(u / q ) ) from four independent frequencies within the sma band based on the q and u maps from the observation .
the q and u values for rm are measured by taking the peak fluxes of the q and u maps , with the image rms uncertainty as the measurement error for q and u. the uncertainty in @xmath0 is calculated from an error propagation of the q and u uncertainties . 0''.8 .
panels clockwise from top left show @xmath48 , @xmath49 , @xmath50 , and @xmath51 . for stokes
@xmath48 , we draw contours at @xmath210 , @xmath25 , 5 , 10 , 40 , 160 , and 480 times the 3 mjy beam@xmath9 rms noise in the image . for @xmath49 , @xmath50 , and @xmath51 , we draw contours at @xmath212 , @xmath29 , @xmath26 , @xmath23 , 3 , 6 , 9 , and 12 times the rms noise in the image , which is 1.3 , 1.3 , and 2.2 mjy beam@xmath9 for @xmath49 , @xmath50 , and @xmath51 , respectively . ] .
the clumps from left to right in the figure are the nucleus ( core ) , jet knot a , b , and c , respectively .
the synthesized beam is shown in the bottom right of the image . ] lccclrl usb ; if 1 & 230.3 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + usb ; if 2 & 232.3 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + lsb ; if 1 & 220.4 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + lsb ; if 2 & 218.4 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + in figure 3 we show the evpas at the four frequencies with 1@xmath53 error bars , together with a fit for the rm in the m 87 core .
the best - fit rm is @xmath2([email protected])@xmath310@xmath4 rad / m@xmath5 .
therefore , we determine the rm to be in the range between -7.5 @xmath3 10@xmath4 and 3.3 @xmath3 10@xmath4
rad / m@xmath5 with a 3@xmath53 confidence level .
since we are interested in an upper limit of the mass accretion rate , which is derived from the upper - limit ( absolute value ) of rm ( see section 4.1 ) , we use 7.5 @xmath3 10@xmath4 rad / m@xmath5 for rm in the following discussion .
and @xmath54 are the inner and outer edge of the faraday screen in the accretion flow where the electrons are sub - relativistic and the magnetic field is coherent . ] to constrain @xmath6 with rm , we essentially follow the procedure outlined in marrone et al .
the primary assumption in this method is that the hot accretion flow in front of a source of synchrotron emission causes faraday rotation .
the model also assumes that the density profile of the accretion flow follows a power - law distribution ( @xmath55 @xmath56 @xmath57 ) and that the magnetic field is well ordered , radial and of equipartition strength . in the case of m87
, we further assume that the _ innermost _ jet ( i.e. its base ) provides the dominant background polarized emission in the submillimeter band ( see figure 4 ) .
note that while polarized emission could be emitted along the jet all the way from close to the bh to radii beyond @xmath12 , the dominant polarized emission at 230 ghz most likely originates from the jet base .
this is because the majority of our sma flux density ( 1.6 jy ) can be explained by the emission from the jet base at the center of m87 .
based on the vlbi observation of m87 at 230 ghz , doeleman et al .
( 2012 ) show that the jet base has a flux density of 1 jy and a size of 5.5@xmath41 . by using the core - shift relationship ( hada et al .
2011 ) , nakamura & asada ( 2013 ) determine the offset of the jet base from the bh at 230 ghz to be 4.3@xmath41 ( in projected distance ) .
therefore , these studies imply that the dominant background emission quite likely originates from within a few @xmath41 from the bh .
nonetheless , we currently can not exclude the possibility of polarized emissions from the diffuse part of the jet , and this issue will be addressed with future high - resolution observations . based on equation ( 9 ) in marrone et al .
( 2006 ) , we express @xmath6 as a function of rm as @xmath58^{-2/3 } \times \left ( \frac{m_{\rm bh}}{6.6\times 10 ^ 9 m_{\odot}}\right)^{4/3 } \left ( \frac{2}{3\beta-1}\right)^{-2/3 } r_{in}^{7/6 } rm^{2/3 } ~\rm rad / m^{2},\ ] ] where @xmath59 is a parameter depending on the accretion flow models ( @xmath59=1/2 @xmath2 3/2 accounts for all varieties of accretion models ) . in the above equation ,
@xmath60 and @xmath61 are the inner and outer edge of the faraday screen ( in units of @xmath27 ) where the electrons are sub - relativistic and the magnetic field is coherent .
we remark that the electrons within @xmath60 make a small or negligible contribution to rm because they become relativistic ( i.e. @xmath62 k=@xmath63 ) and rm is , thus , suppressed ( quataert & gruzinov 2000 ) . in order to derive @xmath6 with the above equation , it is important to determine @xmath60 and @xmath61 .
we estimate @xmath60 based on the observed flux density of m87 from vlbi observations . as we described above , thermal electrons within @xmath60 become relativistic and can emit significant synchrotron radiation that dominates the emission from the accretion disk ( yuan et al .
since it is expected that a hot accretion flow would be optically thick at cm wavelengths , the size of the region with relativistic plasma can be constrained by using the observed flux density and brightness temperature .
for m 87 , the observed total flux density of the nucleus is 1 jy from the vlbi observation at 43 ghz ( abramowski et al .
since the brightness temperature of the hot electrons within @xmath60 must be at least 6@xmath310@xmath64 k , the diameter ( i.e. 2 @xmath60 ) of the emitting region of relativistic plasma can be constrained to be @xmath65 0.32 mas (= 42@xmath27 ) .
since the observed flux density sets an upper limit on the amount of emission from relativistic electrons , it leads to an upper bound on @xmath60 of 21@xmath27 .
on the other hand , if the innermost jet resides right in front of the the emitting region of the relativistic plasma and the emission from the hot electrons is totally absorbed by the optically thick jet , we can not constrain the size of the region of the relativistic plasma directly from the observed flux .
however , we can use the size of the innermost jet to constrain @xmath60 .
the observed size of the vlbi core at 43 ghz is 17 @xmath52 4r@xmath66 ( asada & nakamura 2012 ) . if the emission from the hot electrons is totally absorbed , @xmath60 must be smaller than 17 @xmath52 4r@xmath66 .
therefore , the 21@xmath27 upper bound for @xmath60 is reasonable even for this case .
we fix @xmath61 to be at the bondi radius @xmath12 . here
, @xmath12 is just a convenient value to choose while @xmath61 is not well determined .
this is a reasonable choice because @xmath6 is insensitive to @xmath61 , and @xmath6 will only change by a factor of order unity even if @xmath61 @xmath67 @xmath12 . with these values
, our new measurement of rm sets an upper limit @xmath6 of 9.2 @xmath310@xmath7 m@xmath8 yr@xmath9 at 21@xmath27 for @xmath68 .
this corresponds to 7.4 @xmath3 10@xmath69 @xmath11 , which is at least two orders of magnitude smaller than the bondi accretion rate .
this value implies that the mass accretion rate is significantly suppressed while material is accreted and falling in from @xmath12 to 21@xmath27 .
determining @xmath6 provides an effective way to constrain accretion flow models because each model has its own unique @xmath6 profile .
the adaf model requires @xmath6 to be comparable to @xmath11 from @xmath12 all the way to @xmath70@xmath1@xmath27 , while the adios and the cdaf models require suppression of @xmath6 as ( @xmath71)@xmath72@xmath11 , where @xmath70 is the radius from the bh with 0 @xmath73 @xmath31 @xmath73 1 for adios and @xmath74 for cdaf .
since our new results suggest significant suppression of @xmath6 at the radius of 21@xmath27 , they disfavor the possibility of a classical adaf and prefer adios / cdaf .
this result is in good agreement with the findings from nearly all numerical simulations ( yuan et al .
2012 ) .
the maximum radiative power that can be extracted from mass accretion is @xmath75 @xmath46 @xmath76 @xmath77 7 @xmath310@xmath78 @xmath79 @xmath80 erg s@xmath9 , where @xmath79 is the radiative efficiency . with our upper limit of 9.2 @xmath310@xmath7 m@xmath8
yr@xmath9 (= 7.4 @xmath3 10@xmath69 @xmath11 ) , the maximum radiative power is estimated to be 5 @xmath3 10@xmath81 erg s@xmath9 if @xmath82 .
this is about two orders of magnitude larger than the observed x - ray luminosity @xmath83 of 7@xmath310@xmath84 erg s@xmath9 , suggesting that the current limit of mass accretion rate is sufficient to explain the observed x - ray luminosity . on the other hand ,
the x - ray luminosity is low compared to the maximum available nuclear power .
if @xmath6 is not significantly lower than the upper limit obtained here , this will imply that the radiative efficiency is significantly lower than the canonical value of 0.1 . in this case , both mass accretion rate and radiative efficiency are significantly suppressed which is consistent with the expected properties of riafs .
our upper limit on the rotation measure is derived from a set of polarization position angles ( pa ) with still significant errors .
because of the large uncertainties , it is difficult to demonstrate that pas follow the @xmath85 law , which is the basis of our fundamental assumption on the location of the faraday screen ( relative to the background source ) . as a result , we can not completely rule out the scenario that @xmath86 faraday rotation ( i.e. faraday screen intermixed with emitting plasma in the jet ) plays an important role and the possibility that we are measuring an rm purely originating from cold electrons in the sheath of the jet , especially because the jet in m87 points to the observer nearly along the line of sight ( i.e. @xmath73 20@xmath40 ) . to assess the possibility of a jet - based rm
, we adopt the burn model ( burn 1966 ) on internal faraday rotation and depolarization . by using the degree of linear polarization of the m87 core from the current study ( 1 % ) and that from the optical measurements ( 1%-13% ; perlman et al .
2011 ) , we find that the maximum possible observed rm for the jet at 230 ghz is @xmath16@xmath310@xmath87 rad / m@xmath5 , beyond our current upper limit .
therefore , a jet - originated rm could provide an explanation for our observed rm .
nonetheless , in the context of our simplified one - dimensional model for spherically symmetric riafs , we still prefer the scenario that the faraday rotating plasma ( i.e. accreting gas ) is external to the emission region because the dominant polarized emission quite likely comes from within a few @xmath41 from the bh ( see section 4.1 ) . in this case , since all the non - relativistic electrons reside beyond @xmath121@xmath41 from the black hole , the faraday screen must be external to the background source .
in essence , in order to constrain @xmath6 more unambiguously with the method used in this letter , it is important to establish the @xmath85 law of pa and obtain a more precise rm .
this will rely on higher sensitivity observations with a significantly longer lever arm in the frequency space and higher angular resolution in the future .
the sma is a joint project between the smithsonian astrophysical observatory and the academia sinica institute of astronomy and astrophysics .
we wish to thank all the staff members at the sma who made these observations possible .
abramowski , a. , acero , f. , aharonian , f. , et al .
2012 , , 746 , 151 agol , e. 2000 , apj , 538 , 121 asada , k. , nakamura , m. 2012 , apj , 745 , 28 algaba , j. c. , asada , k. , nakamura , m. 2013 , submitted to apj ; arxiv:1308.5429 blandford , r. d. , & begelman m. c. 1999 , mnras , 303 , 1 bondi , h. 1952 , mnras , 112 , 195 bower , g. c. , et al .
2003 , apj , 588 , 331 burn , b. j. 1966 , mnras , 133 , 67 doeleman et al .
2012 , science , volume 338 , issue 6105 , pp .
355 di matteo , t. , allen , s. w. , fabien , a. c. , wilson , a. s. 2003 , apj , 582 , 133 gebhardt , k. , adams , j. , richstone , d. , lauer , t. r. , faber , s. m. , gltekin , k. , murphy , j. , tremaine , s. 2011,apj , 729 , 119 hada , k. , doi , a. , kino , m. , nagai , h. , hagiwara , y. , kawaguchi , n. 2011 , nature , 477 , 185 ho , p. t. p. , moran , j. , lo , k. y. 2004 , apj , 616 , 1 ichimaru , s. 1977 , apj , 214 , 840 igumenshchev , i. v. & narayan , r. 2002 , apj , 566 , 137 junor , w. ; biretta , j. a. & wardle , j. f. c. 2001 , iaus , 205 , 136 li , y .- r .
, yuan , y .- f . , wang , j .- m . , wang , j .- c .
, & zhang , s. 2009 , , 699 , 513 macchetto , f. , marconi , a. , axon , d. j. , et al .
1997 , , 489 , 579 macquart , j. p. , et al . 2006 , apj , 646 , l111 marrone , d. p. , moran , j. m. , zhao , j. , rao , r. 2006 , apj , 640 , 308 marrone , d. p. , phd thesis , proquest dissertations and theses 2006 .
section 0084 , part 0606 265 pages ; harvard university marrone , d. p. & rao , r. 2008 , spie.7020e , 60 nakamura , m. , & asada , k. 2013 , apj , 775 , 118 narayan , r. , yi , i. 1995 , apj , 444 , 231 narayan , r. , & fabian , a. c. 2011 , mnras , 415 , 3721 narayan , r. , igumenshchev , i. v. , & abramowicz , m. a. 2000 , apj , 539 , 798 owen , f. , hardee , p , e. , cornwell , t. j. 1989 , apj , 340 , 698 owen , f. n. , eilek , j. a. & keel , w. c. 1990 , apj , 362 , 449 pang , b. , et al .
2011 , mnras , 415 , 1228 perlman , e. s. , biretta , j. a. , zhou , f. , sparks , w. b. , & macchetto , f. d. 1999 , , 117 , 2185 perlman et al .
2011 , apj , 743 , 119 quataert , e. , & gruzinov , a. 2000 , apj , 539 , 809 shakura , n. i. , & sunyaev r. a. 1973 , a&a , 24 , 337 yuan , f. , quataert , e. , & narayan , r. 2003 , , 598 , 301 yuan , f. , wu , m. , & bu , d. 2012 , , 761 , 129 walsh , j. l. , barth , a. j. , ho , l. c. , & sarzi , m. 2013 , , 770 , 86 wright , m. c. h. & sault , r. j. 1993 , apj , 402 , 546 zavala , r. t. & taylor , g. b. 2002 , apj , 566 , 9 | we present the first constraint on faraday rotation measure ( rm ) at submillimeter wavelengths for the nucleus of m 87 . by fitting the polarization position angles ( @xmath0 )
observed with the sma at four independent frequencies around @xmath1230 ghz and interpreting the change in @xmath0 as a result of _ external _ faraday rotation associated with accretion flow , we determine the rotation measure of the m 87 core to be between @xmath27.5@xmath310@xmath4 and 3.4@xmath310@xmath4 rad / m@xmath5 . assuming a density profile of the accretion flow that follows a power - law distribution and a magnetic field that is ordered , radial , and has equipartition strength , the limit on the rotation measure constrains the mass accretion rate @xmath6 to be below 9.2@xmath310@xmath7 m@xmath8 yr@xmath9 at a distance of 21 schwarzchild radii from the central black hole .
this value is at least two orders of magnitude smaller than the bondi accretion rate , suggesting significant suppression of the accretion rate in the inner region of the accretion flow .
consequently , our result disfavors the classical _ advection dominated accretion flow _
( adaf ) and prefers the _
adiabatic inflow - outflow solution _ ( adios ) or _ convection - dominated accretion flow _ ( cdaf ) for the hot accretion flow in m 87 . |
spintronics devices require generation of high electron spin polarization in non - resonant excitation conditions and its conservation that must be sufficient to store and manipulate spin information .
we show here that these requirements can be achieved at room temperature in dilute nitride iii - v semiconductor heterostructures under optical pumping . for this
, we use spin - dependent capture of free electrons by deep paramagnetic centers present in semiconductor crystal . the spin - dependent capture results in dynamical polarization of bound electrons which , in its turn , acts as a spin filter increasing polarization of free electrons @xcite .
we demonstrate here , that , after a short optical pump pulse , the decay of both a difference of the free - electron spin - up and spin - down densities and their sum is controlled by the same fast spin relaxation in the conduction band . as a result ,
the spin polarization of free electrons is constant while their total spin decays fast when the time delay increases .
the samples under study are undoped 0.1-@xmath4m - thick gaas@xmath0n@xmath1 ( @xmath2 = 2.1 , 2.7 , 3.5% ) layers grown by mbe on a ( 001 ) gaas substrate @xcite .
we investigate electron spin properties by polarized time - resolved photoluminescence ( pl ) .
the samples are excited along the growth axis ( @xmath5-axis ) by circularly ( @xmath6 ) polarized 1.5 ps pulses generated by a mode - locked ti - sapphire laser with a repetition rate of 80 mhz .
the pl intensities co - polarized @xmath7 and counter - polarized @xmath8 with the excitation laser are recorded in the backward direction using a s1 photocathode hamamatsu streak camera with an overall time - resolution of 8 ps .
we measure the pl circular polarization degree @xmath9 proportional to free - electron spin polarization @xmath10 where @xmath11 and @xmath12 are the densities of spin - up and spin - down free electrons , and @xmath13 is the free - electron total density .
simultaneously , we measure the free - electron ( total ) spin density @xmath14 , which for the bimolecular recombination process writes : @xmath15 , where @xmath16 is the total pl intensity .
n@xmath17 alloy at room temperature .
the laser excitation energy is _ _
h__@xmath18 = 1.39 ev which corresponds to the photogeneration of carriers below gaas barrier .
the pump average power is @xmath19 mw .
detection energy corresponds to the dominant conduction band @xmath20 heavy - hole recombination @xcite .
dashed curves are calculated in framework of the applied model ( see text ) . ]
figure 1 shows the experimental kinetics @xmath21 ( solid line ) , @xmath22 ( points ) and @xmath23 ( circles ) measured in gaas@xmath24n@xmath17 alloy at room temperature @xcite .
one can see that @xmath25 and @xmath26 first rise for @xmath27 ps , which results from relaxation and thermalization of photogenerated carriers . during this time
, @xmath28 varies slightly being close to 25% .
then the pl decays in two steps : the first one is fast and the second one is slower with the characteristic decay times of @xmath29 and @xmath30 ps , respectively . simultaneously with the fast initial pl decay , the pl polarization monotonously rises up to @xmath31% and afterwards keeps the value at least 2 ns as we estimated considering noise fluctuations .
such a long conservation of the high pl polarization is quite surprising since at room temperature spin relaxation time of free electrons in undoped bulk semiconductors of a zinc - bland structure is of the order of magnitude of 100 ps @xcite , i.e. 10 times shorter than we observed .
however , the total electron spin @xmath26 decays fast with a single characteristic time of about 150 ps .
let us give first a qualitative explanation of the effects observed . figure 2 describes schematically two successive regimes of the spin - dependent electron dynamics responsible for the formation of high electron spin polarization and its persistence under the circularly polarized interband pulsed photoexitation .
we assume that ( i ) the density @xmath32 of photoinjected electrons exceeds the concentration @xmath33 of deep centers containing at equilibrium one electron per center , ( ii ) the deep center can only capture a photogenerated electron from the conduction band with a spin antiparallel to the spin of the already present electron , forming a singlet two - electron bound state @xcite , ( iii ) the interband electron - hole recombination contributes negligibly to the balance regulating the electron and hole densities and ( iv ) photoinjected holes are unpolarized because of their fast spin relaxation @xcite .
the first regime starts when all the centers acquire the second electron , the fast electron capture is blocked and the further capture is controlled by the photohole nonradiative recombination with the bound electrons described by the hole lifetime @xmath34 , where @xmath35 is a constant .
thus , the electron and hole densities decay as @xmath36 , @xmath37 ( solid line in fig .
for the pl intensity one has @xmath38 .
this regime ends at the moment @xmath39 when the spin minority photoelectrons disappear from the conduction band : @xmath40 .
this time can be estimated by @xmath41 , where @xmath42 is the initial degree of electron spin polarization @xcite . within the time @xmath39
the minority density exponentially decreases to zero while the density of the majority electrons tends to a finite value @xmath43 which results in a monotonously increasing free - electron spin polarization up to @xmath44 , see the dotted curve in fig .
the second regime is characterized by the almost unoccupied minority spin conduction subband and the spin - polarized bound electrons .
the kinetics is governed by the free - electron spin relaxation time @xmath45 : the recombination rate for the free electrons and holes is given by @xmath46 .
this can be understood taking into account the following balance of rates in the second regime .
the decrease of the electron majority density due to the spin - flip processes is determined by the rate @xmath47 .
the reverse - spin electrons are immediately captured by deep centers which next capture unpolarized holes to generate unpolarized paramagnetic centers .
the majority free electrons are captured at the generated paramagnetic centers with the rate @xmath48 .
this leads to the total decay rate @xmath46 for free photoelectrons , so that @xmath49 , @xmath50 while their spin polarization keeps on at the level @xmath44 .
the densities of one - electron ( paramagnetic ) and two - electron ( nonmagnetic ) centers , @xmath51 and @xmath52 ( @xmath53 ) , respectively , are constant .
particularly , @xmath54 because the hole lifetime equals @xmath45 .
the second regime ends at the moment @xmath55 when the decreasing free - electron density becomes smaller than @xmath52 and the further kinetics is determined by the hole recombination processes .
the time @xmath56 is approximately given by @xmath57 which can remarkably exceed @xmath45 .
it is worth to stress that in contrast to @xmath58 the time @xmath56 is a function of the pump intensity and can be varied in a wide range .
in contrast to the two - stage decrease of the free - electron density @xmath59 with two characteristic times @xmath60 and @xmath45 ( solid line in fig .
2a ) , the spin - dependent difference @xmath61 decays with a single characteristic time @xmath45 ( dashed line in fig .
therefore , the total spin @xmath62 of free electrons is controlled by the short time , @xmath45 , of their spin relaxation : @xmath63 . the experimental data in fig .
1 are obtained at the pump average power of @xmath64 = 130 mw . at this power , as our estimations show ( see below ) , the ratio @xmath65 .
in addition , the pl exponential decay at long delay also indicates bimolecular mechanism of recombination .
this means that @xmath66 and @xmath67 up to 425 ps .
thus , the suggested model can be applied for description of the experiment . since @xmath66 , the gaasn layer under study represents an intrinsic semiconductor where the lifetimes of photogenerated electrons and holes coincide , @xmath68 . at the first stage ,
the pl decay time - constant , @xmath69 , is equal to @xmath70 ps , therefore , @xmath71 ps . at the second stage , where @xmath72 , the pl decay constant equals @xmath73 ps , which yields @xmath74 ps .
n@xmath17 alloy under @xmath6- and @xmath75-polarized excitation ( a ) and under @xmath6 excitation in the absence and presence of a transverse magnetic field ( b ) .
solid and dotted curves are experimental transients , dashed curves are the result of calculation . _ w_(mw ) : 100 ( a ) , 130 ( b ) .
] the pl polarization is not sensitive to free - electron spin relaxation , and , therefore , it can not be used for measuring the time @xmath45 .
however , the total free - electron spin @xmath26 decays exponentially with the time constant @xmath45 . the time was found from the experimental dependency @xmath23 shown in fig . 1 to be @xmath76 ps .
within the experimental error , this value is two times longer than @xmath77 ps which coincides completely with the model predicting @xmath78 .
thus , the measurement of slow pl decay time at spin - dependent recombination allows us to find the free - electron spin relaxation time .
note , that the method is more precise as compared to the measurement of the total spin decay where the ratio of @xmath79 is found .
the suggested model is based on the dynamical polarization of deep centers , which causes a decrease in the recombination rate of the majority spin electrons at the second regime .
if the deep - center polarization is destroyed then the free - electron recombination rate increases up to the one at the fast regime .
the disappearance of the slow range and the pl decay with the single short time @xmath80 ps are observed under the change of the exciting light polarization from the circular ( @xmath6 ) to linear ( @xmath75 ) ( fig .
switching on the perpendicular magnetic field of 7.8 kg under @xmath6-pumping leads to the same effect ( fig .
it again evidences the relevance of the suggested model .
to describe the experimental results quantitatively , we use the set of equations @xcite for the rates of the free- and bound - electron total spins , respectively , @xmath81 and @xmath82 , transformed to the form : @xmath83 here @xmath84 , @xmath85 are the photogeneration rates of electrons with the spin @xmath86 , @xmath87 , @xmath88 is spin relaxation time of bound electrons , @xmath89 the unit vector directed along the growth axis @xmath5 coinciding with the exciting beam , @xmath90 are the larmor frequencies defined by @xmath91 , @xmath92 , @xmath93 and @xmath94 are the @xmath93-factors of free and bound electrons , @xmath95 is the magnetic field . note that @xmath96 , where @xmath97 are the densities of spin - up and spin - down paramagnetic centers , and we assume @xmath98 .
the sign of experimentally observed positive pl polarization in fig .
1 corresponds to the recombination of conduction electrons with heavy holes @xcite . for the recombination , @xmath99 @xcite , where @xmath100 is a depolarization factor .
our model considers kinetics when the pl decay starts .
we assume that the decay begins at the time delay @xmath101 ps which is used as a fitting parameter .
calculated dependencies of the pl intensity @xmath21 , degree of circular polarization @xmath22 and spin @xmath23 shown by dashed curves in fig .
1 and fig . 3a are obtained for the following parameters : @xmath102 and 7.7 for fig . 1 and fig .
3a , respectively , @xmath103 , @xmath104 ps , @xmath88 = 1500 ps , @xmath105 and @xmath106 at @xmath6 and @xmath75 excitations .
one can see , that the calculated curves fit well the experimental ones .
note that the average value of @xmath28 measured within the plateau in fig . 1 equals 43% .
it is two times smaller than the value , 95% , calculated assuming recombination of conduction electrons with heavy holes .
we attribute the difference in the @xmath28 value to admixture of the conduction electron@xmath20light - hole recombination having the opposite sign of @xmath28 @xcite . a coincidence of the calculated ( dashed ) curve @xmath107 with the experimental one in fig .
1 is obtained using the fitting coefficient of @xmath108 .
the calculations ( not presented here ) show that the normalized relations @xmath107 and @xmath109 depend weakly on variation of the parameters @xmath110 and @xmath111 . specifically , they virtually coincide at @xmath112 and @xmath113 . at the same time
, we found that calculated for a continuous - wave ( cw ) excitation dependencies @xmath114 and @xmath28 on both the excitation intensity and transverse magnetic field ( hanle effect ) are very sensitive to these parameters .
fitting of experimental curves found in ref .
[ @xcite ] at cw excitation of the same gaas@xmath24n@xmath17 sample allowed us to estimate the value of @xmath115@xmath116 .
this provided in turn to find the relation of @xmath117 and @xmath118 at the average pulse pump power @xmath119 mw .
the value @xmath120 ps used in this paper is in agreement with the 600 ps estimate from below for @xmath88 found in ref .
[ @xcite ] from the analysis of the hanle effect .
it should , however , be mentioned that , in comparison to the continuous - wave measurements of ref .
[ @xcite ] , the time - resolved data are less sensitive to @xmath88 .
particularly , the `` plateau '' of the curve @xmath107 lasts out up to 500 ps even for @xmath88 decreased to 30 ps which clearly demonstrates dynamical character of the observed spin - dependent effects .
4 shows the beats in the @xmath122 ( solid ) and @xmath123 ( dashed ) pl components and in the polarization degree @xmath28 ( circles ) recorded in gaas@xmath124n@xmath17 alloy in the presence of the transverse magnetic field @xmath125 kg . these beats result from the larmor precession of electron spins .
they have a complex shape since @xmath93-factors of free and bound electrons , @xmath93 and @xmath94 , differ both in their value and sign . however , the exact values of @xmath93 and @xmath94 in gaas@xmath124n@xmath17 are unknown . since in our measurements @xcite the sign of @xmath94 was found to be positive ,
we believe that the @xmath93-factor of bound electrons is close to the @xmath93-factor of electrons in vacuum , i.e. , @xmath126 . as for @xmath93-factor of free electrons , a paper devoted to measuring @xmath93 in gainasn dilute nitrides
is known only @xcite .
it was found in this paper that @xmath93 is negative .
we used @xmath93-factor @xmath93 = @xmath200.9 in our calculations since at this value the calculated beat period fits the experimental one .
it is seen in fig . 3b
that the peak @xmath21 is delayed by 14 ps in the magnetic field 7.8 kg as compared to the case of _ _ b__=0 .
in accordance to this , we assume that , in the presence of magnetic field , @xmath127 ps .
dotted line in fig .
4 shows the beats of @xmath28 calculated with @xmath93 = @xmath200.9 , @xmath128 and @xmath127 ps .
one can see that our model describes qualitatively the experimental beats for @xmath129 ps .
we have also observed strong pl polarization and its stability at room temperature in an @xmath130-doped ( @xmath131@xmath116 ) gaasn layer and in an undoped compressively strained quantum well ingaasn / gaas with small nitrogen content of the order of 1% @xcite .
this indicates the general character of the observed effects . to conclude
, we have measured the giant polarization ( @xmath132 ) of free electrons and its persistence ( @xmath133 ns ) in gaasn alloys at room temperature .
we have developed the nonlinear theory of spin dynamics in a coupled system of free and localized carriers controlled by spin - dependent recombination .
the theory shows that the increase of both free and localized electron polarizations up to their limiting values is due to dynamical polarization of deep paramagnetic centers .
when the latter is effective , the mean spin of free electrons is independent of their spin relaxation while the total free - electron spin decays in the conduction band with the short spin relaxation time of @xmath134 ps .
also , electron spin quantum beats are observed in transverse magnetic field at room temperature and described theoretically . | we report on both experimental and theoretical study of conduction - electron spin polarization dynamics achieved by pulsed optical pumping at room temperature in gaas@xmath0n@xmath1 alloys with a small nitrogen content ( @xmath2 = 2.1 , 2.7 , 3.4% ) .
it is found that the photoluminescence circular polarization determined by the mean spin of free electrons reaches 40 - 45% and this giant value persists within 2 ns .
simultaneously , the total free - electron spin decays rapidly with the characteristic time @xmath3 ps . the results are explained by spin - dependent capture of free conduction electrons on deep paramagnetic centers resulting in dynamical polarization of bound electrons .
we have developed a nonlinear theory of spin dynamics in the coupled system of spin - polarized free and localized carriers which describes the experimental dependencies , in particular , electron spin quantum beats observed in a transverse magnetic field . |
extensive deep multi - wavelength observations across the sky have vastly improved our knowledge of star - formation and agn activity across a range of redshifts .
it has become abundantly clear that the universe at @xmath6 , was much more active than it is currently , at @xmath7 .
for example , the star - formation rate density steadily rises from @xmath8 to @xmath9 , flattens out until at least @xmath103 and then falling off again at the highest redshifts @xcite .
agn activity shows a similar form , indicating that globally , at least , the growth of supermassive black holes roughly tracks the buildup of the stellar population in galaxies ( e.g. @xcite ; @xcite ; @xcite ) .
observing galaxies at z@xmath112 should give us tremendous insight into this evolution , and the interplay between agn and star - formation in their host galaxies during this more active phase , but spectroscopic identifications , and hence our understanding of galaxies with @xmath12 is complicated by the redshift desert " .
at these redshifts there are very few spectral features at observed - frame optical wavelengths that can be used to determine the redshift . in response to this problem
numerous selection techniques have been developed that use broad band photometry instead of spectroscopy to select @xmath102 candidate galaxies .
for example , the lyman - break and balmer break features have been used to select galaxies with @xmath13 @xcite .
the bzk selection technique of @xcite combines optical ( @xmath14 and @xmath15 band ) and near - ir ( @xmath16 band ) photometry to target the balmer and 4000 break features of star - forming and passively evolving galaxies at @xmath102 @xcite .
the selection technique is expected to be largely independent of reddening @xcite and the near - ir photometry allows extremely red passively evolving galaxies to be selected , as well as bluer , star - forming galaxies @xcite .
bzk selection shows good agreement with other @xmath102 galaxy selection techniques , although it is liable to miss the bluest / youngest star - forming galaxies ( k@xmath17 vega mag ; @xcite ) .
many sbzks show strong emission in the mid - ir at 24@xmath1 m , the majority of which is attributable to dusty star formation @xcite . by comparing uv and ir derived star - formation rates ( sfr s )
@xcite identified sbzks with excess ir emission ( ir excess sbzk ) as candidates for containing agn .
@xcite to attempted to isolate sbzks which might contain agn using this mid - ir excess .
agn are expected to exhibit strong continuum emission at mid - ir wavelengths due to re - radiated emission from hot dust ( e.g. @xcite ; @xcite ) , a feature that has been exploited previously to identify obscured agn @xcite .
@xcite stacked the x - ray emission of ir excess sbzks in the goods - s field , yielding a hard x - ray signal ( @xmath18 ) consistent with a significant population of compton thick ( n@xmath19 @xmath20 ) agn .
the stacked signal of x - ray undetected sbzks that did not exhibit excessive ir emission ( ir non - excess sbzks ) appeared softer ( @xmath21 ) and consistent with pure star - formation .
these results have significant implications for co - eval star - formation and black hole growth and the composition of the x - ray background .
the presence of buried " compton thick agn in star - forming galaxies ( like sbzks ) has been predicted by models of agn - galaxy - co - evolution ( e.g. @xcite ) .
the majority of the x - ray background @xcite has been resolved using the emission of discrete x - ray sources ( as predicted by @xcite ) , but there remains an unexplained hard ( e @xmath22 6 kev ) x - ray deficit @xcite .
models of the x - ray background predict the hard x - ray background deficit could represent a large population of as yet undiscovered compton thick agn at @xmath23 @xcite . extrapolating the stacked x - ray emission of the x - ray undetected ir - excess sbzks over all
the sky would explain a substantial portion ( @xmath24 ) of the hard x - ray background deficit @xcite .
@xcite carried out further analysis of sbzks using new , deeper 4ms cdfs x - ray data and found that stacking x - ray spectra of heavily obscured ( @xmath25 ) sbzks produced a hard , reflection dominated spectrum consistent with compton thick agn emission .
the stacked spectrum features a large ew iron line at rest - frame e @xmath11 6.4 kev similar to that observed in compton thick agn in the local universe @xcite and distant universe @xcite .
stacked the spectra of sbzk galaxies containing heavily obscured agn and found their spectra was best fit by a pure reflection model with a strong iron k@xmath26 line at rest - frame 6.4 kev .
this is consistent with compton thick absorption ( n@xmath27@xmath28 ) in a small fraction of sbzk galaxies . stacking x - ray undetected sbzks in deeper 4ms cdfs data
did not yield conclusive evidence of a high fraction of heavily obscured agn in the more general sbzk population , however , somewhat in disagreement with the earlier work of @xcite .
these observations imply that the space density of compton thick agn at @xmath0 is comparable to that of unobscured agn at @xmath0 .
the ir excess analysis of @xcite is reliant upon the accuracy of ir and uv sfr estimates .
mid - ir observations have been found to be reliable proxies for bolometric ir luminosities @xcite and it has been common practice to fit 24@xmath1 m observations ( rest - frame 8@xmath1 m at @xmath0 ) with empirically derived star - forming galaxy templates ( e.g. @xcite ; @xcite ) to estimate the sfr . with the advent of _
herschel _ our understanding of the far - ir spectra of star - forming galaxies has improved markedly , especially at @xmath29 .
using far - ir _ herschel _ data @xcite calculated @xmath5 for numerous galaxies with mips coverage and compared them to @xmath5 estimates made using 24@xmath1 m selected star - forming templates of @xcite .
based on this analysis it is concluded that the star - forming templates of @xcite are overestimating the bolometric ir luminosity for galaxies with @xmath30 @xmath31 . additionally @xcite find that the ir excess emission at rest - frame 8@xmath1 m can be explained entirely by pah features , removing the necessity for mid - ir continuum emission from agn assumed by previous ir selection techniques .
the uv estimates of sfr are known to suffer in dusty ( ie heavily star - forming ) galaxies because reddening corrections become less accurate as the dust opacity increases @xcite .
thus , selection of agn via comparison of ir and uv - derived sfrs may not be as robust as previously thought .
indeed , recent work have suggested that some ir excess selection techniques ( e.g. @xcite ) have been shown to select equal parts heavily obscured agn , low - luminosity agn and purely star - forming galaxies @xcite when followed up using x - ray data . in the light of this ,
in the present paper we revisit the issue of the agn content in ir excess bzk galaxies pioneered by @xcite .
we examine the x - ray properties of ir luminous sbzks , selected in a similar manner to @xcite , using 4ms cdfs and 2ms cdfn x - ray data , through a mixture of direct detections and stacking . in section [ sub : optir ]
we discuss the optical and ir data used in this work .
section [ sub : sample ] outlines the bzk and 24@xmath1 m selection methodology while sections [ sub : xray_redux ] and [ sub : stacking ] outline the x - ray data reduction and x - ray stacking technique respectively . in section [ sec : results ] we present the results of direct x - ray detection of sbzks and stacking of x - ray undetected sbzks and in section [ sec : comparison ] we compare our findings to those of @xcite and @xcite . in section [ sec : discussion ]
we discuss the implications of these results and in section [ sec : summary ] we summarise our findings . throughout this work a standard , flat @xmath32cdm cosmology with @xmath33 and h@xmath34 km s@xmath35 mpc@xmath35 is assumed .
in the cdfs we use _ hubble _ acs f435w , f606w , f775w and f850lp band ( henceforth @xmath36 and @xmath15 respectively ) , isaac @xmath16 band , irac 3.6@xmath1 m and mips 24@xmath1 m band data from the goods - music - v2 catalogue ( @xcite ) .
all objects have either a spectroscopic ( 64% ) or a photometric ( 36% ) redshift .
this dataset presents identical @xmath37 and 3.6@xmath1 m photometry to the goods - music catalogue ( @xcite ) that was used by @xcite but incorporates deeper mips 24@xmath1 m band data from the fidel survey ( pi : mark dickinson ) .
the 24@xmath1 m source fluxes in goods - music - v2 were identified and matched to the multi - wavelength data using the convphot software @xcite , as opposed to matching the mips sources to multi - wavelength data using a more basic fixed matching radius method ( as was the case for @xcite and subsequently daddi et al 2007a and 2007b ) .
the additional depth and use of the convphot software allows the identification of additional 24@xmath1 m sources and therefore increases our sample size compared to @xcite .
the goods - music - v2 catalogue has more spectroscopic redshifts than goods - music , improving the accuracy of our luminosity calculations . in the cdfn
we use _ hubble _ acs @xmath38 and @xmath15 band , moircs @xmath39 , irac 3.6@xmath1 m and mips 24@xmath1 m band photometry collected for the moircs deep survey ( mods ; @xcite ) .
mods is the only publicly available catalogue in cdfn to date to have matched k band to 24@xmath1 m sources .
again , sources have a mixture of spectroscopic ( 70% ) and photometric ( 30% ) redshifts . 1 . a secure k - band detection ( @xmath40 vega ) .
2 . less than 0.5 arcseconds separation from an irac 3.6um selected source to verify that the source is genuine ( large psfs of 24@xmath1 m sources leads to source blending ) .
3 . error on @xmath41 to ensure reddening corrections are reliable .
4 . either a spectroscopic or photometric redshift . | we investigate the x - ray properties of bzk - selected galaxies at @xmath0 using deep x - ray data in the chandra deep field south ( cdfs ) and chandra deep field north ( cdfn ) .
a subset of these bzk galaxies have been proposed as compton thick active galactic nuclei ( agn ) candidates based on a high ratio of infrared ( ir ) and ultraviolet ( uv ) star formation rates ( sfr ) . with the benefit of deep 24@xmath1 m observations
our sample of these ir excess galaxies is larger than previous studies , and combined with the deepest x - ray data yield new insights into the population .
we identify 701 and 534 star forming bzks ( sbzks ) in the range @xmath2 in the cdfs and cdfn respectively . of these
we directly detect in x - rays 49 sbzks in cdfs and 32 sbzks in cdfn . stacking the undetected sources also reveals a significant x - ray signal . investigating the x - ray detection rate and stacked flux versus the ir excess parameter ( i.e. sfr@xmath3/sfr@xmath4 ) , we find no strong evidence for an increased x - ray detection rate , or a harder x - ray spectrum in ir excess sbzks .
this is particularly the case when one accounts for the strong correlation between the ir excess parameter and the bolometric ir luminosity ( @xmath5 ) , e.g. when controlling for @xmath5 , the ir non - excess sbzks show a detection rate at least as high .
while both direct detections and stacking suggest that the agn fraction in sbzk galaxies is high , there is no clear evidence for widespread compton thick activity in either the sbzk population generally , or the ir excess sbzk subsample .
the very hard x - ray signal obtained for the latter in earlier work was most likely contaminated by a few hard x - ray sources now directly detected in deeper x - ray data .
the x - ray detection fraction of passive bzk galaxies in our sample is if anything higher than that of sbzks , so there is no evidence for coeval black hole growth and star formation from x - ray analysis of the bzk populations . because increased agn activity in the ir excess population is not indicated by our x - ray analysis , it appears that the bulk of the ir excess sbzk population are luminous star - forming galaxies whose sfrs are either overestimated at 24@xmath1 m , underestimated in the uv , or both .
this conclusion reinforces recent results from _ herschel _ which show similar effects .
catalogues - galaxies : active - galaxies : nuclei - galaxies : star formation - infrared : galaxies - ultraviolet : galaxies - x - rays : galaxies |
a supernova ( sn ) event in our own galaxy will lead to a large flux of neutrinos arriving at the earth within the space of a few seconds .
the only such event detected so far has been supernova 1987a @xcite , at a time when neutrino detectors were considerably smaller and less advanced than they are today .
hence the detection of another such event , which should occur within the near future , will result in a great deal of new information regarding both supernovae and neutrinos @xcite .
the majority of these neutrinos are produced thermally during the supernova event , where temperatures can reach around 10 mev @xcite , and carry away most of the released energy .
they are emitted with a distribution with is approximately of the fermi - dirac form , with a temperature which differs for the neutrino flavours .
broadly the electron neutrinos should have the lowest temperature as they scatter the most in the star and so decouple at the largest radius .
by contrast electron anti - neutrinos scatter less with the stellar matter and so are emitted at a higher temperature , and the remaining heavy flavours have a higher temperature still .
various authors have demonstrated that , in addition to neutrinos , other more exotic particles could be produced thermally during the supernova event @xcite .
these particles @xmath0 are ` dark ' in the sense that they interact only weakly with visible matter and so are difficult to detect , except perhaps in extreme environments such as a hot post - supernova star .
indeed if such particles interact with neutrinos through scattering or annihilation they will alter the various temperatures of the neutrino spectrum @xcite , which could result in observable effects on earth . however the exact temperatures are not well known and depend upon assumptions related to the structure of the star during thermal production , specifically the dependence of density and temperature with radius @xcite .
in this work we seek to understand how the scattering with neutrinos of additional thermally - produced mev - mass particles will alter the neutrino temperatures , given the potentially large uncertainties in the modelling of the star during the supernova event .
using this information we quantify to what extent current and future experiments can exploit measurements of the neutrino temperatures to constrain the existence of such particles .
we focus specifically on the difference between the temperatures of the neutrino flavours , since dark sector scattering with neutrinos drives the temperatures of the different flavours towards a single value .
this is important since , as shown in figure [ fig : spectra ] , a spectrum which is the sum of two ( or more ) different temperatures is clearly distinct from one which is made up of only one temperature . from a supernova ,
whose spectrum is a combination of fermi - dirac distributions with two temperatures at 5 mev and 8 mev .
we compare this with spectra with only a single temperature of either 5 mev and 8 mev and the combined spectrum.,scaledwidth=49.0% ] in section [ sec : production ] we outline how the temperatures of the neutrino species are calculated and in section [ sec : chi ] we describe how these are modified in the presence of new states @xmath0 . in section [ sec : detection ] we give the formulae for the neutrino flux spectrum at earth from a galactic supernova and how this is converted to the spectrum in a detector .
we use these formulae in section [ sec : cons ] to set projected upper limits on the scattering between @xmath0 and neutrinos using a bayesian analysis and compare these to other constraints in section [ sec : comparison ] .
we conclude in section [ sec : conc ] .
in this section we outline how the different neutrino temperatures @xmath4 and number densities @xmath5 are calculated using the diffusion equation , followed in section [ sec : chi ] where we describe how these are affected by scattering with a new ` dark ' particle @xmath0 .
our approach is approximate , and we do not claim accuracy on par with numerical simulations of the supernova event @xcite , however it is sufficient for our purpose of understanding how sensitive the different neutrino temperatures are to modelling uncertainties in the star during thermal production .
the diffusion equation takes the form @xcite , @xmath6 where @xmath7 is the radial distance from the centre of the star , a prime denotes an @xmath7-derivative , @xmath8 is the diffusion coefficient , @xmath9 is the self - annihilation cross section for the particle , @xmath10 is the number density of the particle when it is in thermal equilibrium with the stellar matter , @xmath11 is the lorentz factor and @xmath12 is the lifetime of the particle . the value of @xmath7 where @xmath5 departs from its equilibrium value i.e. where @xmath13 is called the neutrinosphere radius ( in the case of neutrinos at least ) .
this is effectively the radius where the neutrinos can be said to have decoupled from the stellar matter . likewise
the corresponding temperature @xmath4 of the star at this radius is the neutrinosphere temperature .
the neutrinosphere temperature determines the observed neutrino temperature on earth from the supernova .
this can be approximated by reducing equation to the following expression @xcite , @xmath14 and solving for @xmath4 , where we assume that @xmath10 is , @xmath15 \mathrm{exp } \left(\frac{- m}{t(r ) } \right ) , \label{eqn : equib}\ ] ] where @xmath16 is the mass of the particle ( e.g. a neutrino or dark particle @xmath0 ) .
we approximate the radial dependence of the baryon density and temperature of the star after the supernova event using a simple empirical model , whose parameters we can vary easily to understand the effect of their uncertainty on the different neutrino flavour temperatures .
as such we assume the following power - law forms for the baryon density @xmath17 and the baryon temperature @xmath4 , @xmath18 where @xmath19 , @xmath20 , @xmath21 , @xmath22 and @xmath23 are in principle all free parameters .
some particular models take for example : @xmath24 , @xmath25 , @xmath26 mev , @xmath27 km and @xmath28 @xmath29 @xcite ( model a ) or @xmath30 , @xmath31 , @xmath32 mev , @xmath33 km and @xmath34 @xmath29 @xcite ( model b ) . for our projected limits we will treat these variables as nuisance parameters , since the exact form of the profile
is not known and can vary significantly .
the value of the neutrinosphere temperature is fixed by the scattering cross section between neutrinos and the other particles of the star , particularly baryons , and the self - annihilation cross section of the neutrinos .
the former effect arises through the diffusion coefficient , which if we assume dominant scattering from baryons equals @xmath35 @xcite , where @xmath36 is the velocity of the neutrinos , @xmath37 is the scattering cross section with baryons through the channel labelled by @xmath38 ( e.g. neutral - current ) and @xmath39 is a factor relating to the relative abundance of neutrons and protons .
electron neutrinos scatter the most frequently and so have the smallest value of @xmath8 due to their interaction with neutrons via the charged current interaction as well as the sub - dominant neutral current channel .
electron anti - neutrinos by contrast scatter with protons via charged current interactions instead , whose density is several orders of magnitude smaller than neutrons in the star , resulting in a larger value of @xmath8 .
the remaining ` heavy ' flavours have the largest @xmath8 and so the largest neutrinosphere temperature as they dominantly scatter with neutrons and protons via neutral current interactions , whose cross section is smaller than that for charged - current scattering .
this all holds to first order , however effects from more detailed modelling of the supernova can alter this simple picture @xcite .
we now consider the possibility that an additional species of particle @xmath0 , which interacts only weakly with the visible sector , is produced thermally in the supernova and scatters with neutrinos , thereby contributing to the diffusion coefficient @xmath8 . under the assumption that such new interactions are flavour - blind the diffusion coefficient becomes , @xmath40^{-1 } , \label{eqn : diff_chi}\ ] ] where @xmath41 is the number density of the new species @xmath0 and @xmath42 is its scattering cross section with neutrinos .
importantly if @xmath43 then the value of @xmath8 will be dominated by the flavour - blind scattering with the dark sector @xmath0 , and so we might expect the neutrino temperatures to approach a single value regardless of flavour .
the exact temperature dependence of @xmath42 depends on how the @xmath0 particles interact with neutrinos .
for example following refs .
@xcite if one assumes either scalar @xmath0 interacting with neutrinos via the exchange of a new fermion , or fermionic @xmath0 interacting with neutrinos via a new gauge boson , then @xmath44 is a reasonable assumption provided that the masses of @xmath0 and the exchange particle are not degenerate .
for the former case of scalar @xmath0 we envisage a term in the lagrangian of form @xmath45 where @xmath46 is a heavy right - handed fermion which mediates the interaction between @xmath0 and neutrinos .
this then leads to a thermally averaged scattering cross section of form @xmath47 .
such an interaction term could also result in the generation of a mass term for the neutrinos in analogue to the see - saw mechanism ( see e.g. ref .
hence our projected constraints are partly motivated as a method of constraining these neutrino mass models , however they can be interpreted in the context of other models and we have checked our results in the case of alternative assumptions , such as @xmath48 , and find no significant deviation from the conclusions presented here .
furthermore the form of @xmath41 depends on the interactions of the @xmath0 particle with nucleons and electrons as well as neutrinos . in this work
we assume that the @xmath0 are in thermal equilibrium up to a radius larger than the neutrinosphere for all flavours .
this means that @xmath0 has either a large annihilation cross section or decay rate and/or interactions with baryons in order to keep the @xmath0 coupled to the hot stellar matter .
we return to this assumption in section [ sec : comparison ] .
we restrict our analysis to particles with mass @xmath49 mev to avoid constraints from nucleosynthesis on thermally produced light particles @xcite .
this also means that the density of @xmath0 will be boltzmann suppressed compared to that for neutrinos , meaning that the latter still dominate the energy transport from the supernova ( see section [ sec : comparison ] for further discussion ) .
( of form @xmath50 ) with a mass of 10 mev .
the dotted line represents @xmath51 , the dashed represents @xmath52 and the solid line is for the remaining heavy flavours .
we show such dependence for two different assumptions on the profile of the matter in the star during the supernova event , labelled as model a and model b.,scaledwidth=49.0% ] we show in figure [ fig : temps_plot ] the values of the three different neutrino temperatures as a function of @xmath42 for the two different example empirical models of the star introduced earlier as model a and model b. these have been calculated using equation ( [ eqn : thermal_decoupling ] ) with @xmath8 from equation ( [ eqn : diff_chi ] ) .
there are two main points to note here : the first is that the values of the neutrino temperatures depend strongly on the choice of empirical model for the star during the supernova event .
the second is that when @xmath42 is large enough the values of the temperatures for each flavour become equal , and that this value of @xmath42 is more robust to changes in the empirical model of @xmath53 and @xmath54 than the absolute temperatures themselves .
therefore if we are to obtain anything useful from analysing the temperatures of neutrinos from a supernova it will likely come from comparing the relative temperatures between species , rather than their absolute values .
_ hence although the values of the temperatures depend strongly on the model of the hot star for a given value of @xmath42 , the number of distinct temperatures which make up the total neutrino spectrum is much less sensitive . _ in section [ sec : cons ] we use this fact to understand what can be learned in future detectors by determining the number of different temperatures in the overall spectrum .
in this section we outline how neutrinos from a supernova will be detected on earth using either inverse - beta decay capture of electron anti - neutrinos @xcite in experiments such as super - kamiokande @xcite or hyper - kamiokande @xcite or via coherent nuclear recoils in direct detection experiments @xcite .
the former has the advantage of greater statistics , while the latter is insensitive to the mixing between neutrino flavours on their way to earth and allows for all three temperatures to in principle be measured .
the expected spectrum of neutrinos @xmath55 from a supernova takes the form of a pinched fermi - dirac distribution @xcite ( neglecting features from mixing effects @xcite ) , @xmath56 where @xmath57 is the temperature of the particular neutrino species ( of which there are a number @xmath58 ) , @xmath59 is the relative luminosity of species @xmath38 compared to the total , @xmath60 is the pinching parameter , which we assume to be zero in this work for simplicity , and @xmath61 is a dirac integral , taking the form @xmath62 multiplying @xmath63 by the total flux from a galactic supernova @xmath64 gives the flux spectrum on earth .
we estimate the total flux using the time - integrated supernova luminosity @xmath65 ergs @xcite , the average energy of each flavour @xmath66 and the distance to the supernova @xmath67 kpc ( i.e. close to the galactic centre ) .
hence the total integrated flux spectrum of neutrinos from a galactic supernova takes the form , @xmath68 detectors such as super - kamiokande and hyper - kamiokande @xcite will detect supernova neutrinos primarily via inverse - beta decay ( i.e. @xmath69 ) due to the large cross section for this process @xcite .
the events are identified by positrons whose energies @xmath70 are related simply to the neutrino energies @xmath71 as @xmath72 .
the differential detection rate is then the product of @xmath73 with the cross section for inverse beta decay .
however the calculation is complicated by the fact that we only detect @xmath52 which will be made up of neutrinos which were in various different flavours at the point of production .
hence the spectrum of the @xmath52 at earth should be composed of two distinct temperatures provided that @xmath42 is negligible and that the different anti - flavours mix on their way to earth .
this also means that the temperature of the @xmath51 flavour is not measurable with this detection method , and so there will only at most be two temperatures .
hence when we determine projected limits on @xmath42 we need to take account of the potentially uncertain values of the relative luminosities @xmath59 for the two remaining temperatures .
by contrast , the detection of neutral current scattering allows all of the neutrino temperatures to be detected regardless how how the neutrinos mix on their way to earth .
direct detection experiments looking for dark matter could be sensitive to supernova neutrinos via this channel . in this case
the neutrinos scatter coherently on nuclei with a differential cross section @xmath74 @xcite leading to a differential scattering rate of nuclear recoil events taking the form , @xmath75 where @xmath76 is the minimum neutrino energy able to induce a recoil of energy @xmath70 for a nucleus of mass @xmath77 . in this case
since the detection method is via neutral - current scattering all three temperatures can be detected and the relative luminosities @xmath59 are set at the point of production ( ignoring possible new physics effects during transit ) . with dark sector particles @xmath0 in an experiment detecting neutrinos via inverse beta decay ( similar to super kamiokande @xcite ) or by nuclear scattering .
the different spectra correspond to the neutrino flavour temperatures predicted assuming model a in figure [ fig : temps_plot].,title="fig:",scaledwidth=49.0% ] + with dark sector particles @xmath0 in an experiment detecting neutrinos via inverse beta decay ( similar to super kamiokande @xcite ) or by nuclear scattering .
the different spectra correspond to the neutrino flavour temperatures predicted assuming model a in figure [ fig : temps_plot].,title="fig:",scaledwidth=49.0% ] shown in figure [ fig : spectra_sigmas ] are the expected spectra in both types of detector considered here , for different values of @xmath42 .
as @xmath42 increases the spectra shift towards having a single temperature as opposed to being composed of multiple distinct temperatures . this is most obvious with knowledge of the whole spectrum , and so in order to discriminate between the different spectra we need to be able to observe both low and high energies events , and so a low threshold is crucial .
we are now in a position to derive projected exclusion limits on the dark sector interaction cross section with neutrinos @xmath42 , using the spectra calculated in this section .
how strong these limits can be will be determined by how effectively we can discriminate the different constituent temperatures in the spectrum from the supernova , and the systematic uncertainties from the modelling of the star during the supernova event .
in this section we determine the projected sensitivity of future multi - tonne experiments to the scattering cross section between dark sector particles and neutrinos , produced thermally in a galactic supernova .
such interactions will alter the temperatures of the different neutrino flavours ( see figure [ fig : temps_plot ] ) , however exactly how this happens depends crucially on how well we know the dependence of the baryon density and temperature of the star during the supernova event .
hence our projections must incorporate this uncertainty in the choice of profile . to do this we perform a bayesian analysis , while parametrising the empirical model of the hot star with nuisance parameters , which we marginalise over .
for such parameters we use @xmath23 , @xmath22 and @xmath21 from equations ( [ eqn : nbprof ] ) and ( [ eqn : tprof ] ) , which determine how steeply the temperature and baryon density fall off with radius . the posterior density , used to determine the exclusion sensitivity for a particular data - set @xmath78 , is @xmath79 where @xmath80 represents the likelihood function , which we assume to be of poisson form to compare the expected spectrum for a given stellar profile and value of @xmath42 ( e.g. figure [ fig : spectra_sigmas ] ) to simulated data .
@xmath81 are the prior distributions for the three nuisance parameters and the signal parameter @xmath42 .
we then set a projected limit by generating a number of simulated data - sets @xmath82 given a particular detector mass and use the posterior summed over these data - sets @xmath83 .
we generate these data - sets by assuming a particular set of temperatures and relative luminosities for each flavour .
integrating @xmath84 from @xmath85 up to @xmath3 of the total posterior probability then gives us our projected exclusion limits .
the choice of priors for the nuisance parameters represents how well these quantities , and thus the profile of the baryonic matter during neutrino ( and @xmath0 ) thermal production , are known .
since we are interested in understanding how well @xmath42 can be probed even if the stellar profile is poorly known we choose priors which cover a wide - range of @xmath22 , @xmath23 and @xmath21 . in principle
however the profile uncertainty could be reduced using e.g. simulations of the supernova event @xcite .
figure [ fig : lims ] shows projected limits for either detection via inverse - beta decay or coherent nuclear scattering ( other detection methods have also been proposed e.g. ref .
@xcite ) assuming priors which equal a constant in the following ranges ( and zero otherwise ) , @xmath86 \\
\mathcal{p}(q ) & \in & [ 1,10 ] \\
\mathcal{p}(t_0 ) & \in & [ 2,20 ] \ ,
\mathrm{mev},\end{aligned}\ ] ] with @xmath20 and @xmath19 fixed at values of @xmath27 km and @xmath28 @xmath29 .
the choice of these priors is arbitrary , and we have intentionally chosen them to represent a large degree of uncertainty by spanning a wide range of possible stellar profiles , including the values of @xmath23 , @xmath22 and @xmath21 from model a and model b introduced earlier .
greater knowledge of the profile of the star during neutrino production would reduce the range of these priors and so would likely strengthen our projected limits . in order to determine these projected limits we have simulated datasets in detectors of various target masses . since we do not know _ a priori _
what spectra they will observe , depending on the physics during the supernova and the mixing of the neutrino flavours , we have set limits under two different assumptions .
for example in the left panel of figure [ fig : lims ] we set limits under the assumption that the spectrum of @xmath52 from the supernova event is either made up of a single fermi - dirac spectrum of temperature 5 mev or of two spectra with temperatures of 5 mev and 8 mev . for the former case we do not know whether @xmath0-neutrino scattering has forced the temperatures of the different anti - neutrino flavours together , whether a different mechanism has brought the temperatures together , or whether the electron anti - neutrino flux is dominated by neutrinos which were produced in that same flavour at the supernova . hence conservatively all one can do is set an upper limit on @xmath42 at the level where it would drive the temperature of the neutrinos
to be smaller than 5 mev , given uncertainties on the stellar profile .
however by comparison the projected limit on @xmath42 will be much stronger if a spectrum with a combination of temperatures is observed ( see also figure [ fig : spectra ] ) , which is the more likely scenario if the neutrino flavours mix on their way to earth . in this case
as can be seen from the left panel of figure [ fig : lims ] the exclusion limit will be at the level of @xmath1 @xmath2 for super kamiokande @xcite , assuming two temperatures at 5 mev and 8 mev with equal relative luminosities .
this is partly due to it being easier to discriminate different spectra with information on multiple temperatures , given the large uncertainties in the profile of the baryons after the supernova event .
essentially with enough statistics a spectrum with two temperatures can be easily distinguished from one with only one temperature , even if the uncertainties on the values of the temperatures themselves are significant .
the limit which hyper kamiokande @xcite should be able to set should be at least an order of magnitude stronger due to the larger number of events it will be able to observe .
we show in the right panel of figure [ fig : lims ] the projected limits for a xenon - based experiment which observes supernova neutrinos through coherent nuclear scattering with a combined spectrum of three temperatures at 8 mev , 5 mev and 3.5 mev , similar to the result for model a in figure [ fig : temps_plot ] . in this case
there is no uncertainty from the mixing of the different neutrino flavours since the detection is via the neutral - current , however the smaller target masses of experiments which exploit this effect , mainly dark matter direct detection experiments , means that they are unlikely to build up enough statistics to compete with super or hyper kamiokande .
nevertheless a xenon - based detector with a mass around 20 tonnes ( e.g. darwin @xcite ) should be able to set a limit on the dark sector scattering cross section with neutrinos provided the threshold for nuclear recoils can be lowered to around 0.1 kev
. this may be possible if such detectors look only at ionisation signals , for which the threshold is generally lower , since discrimination between nuclear and electronic recoils will not be required for such a significant event burst .
the key advantage of the direct detection experiments is their ability to measure all three potential temperatures , while relying on inverse beta decay capture of @xmath52 allows only the temperatures of the anti - flavours to be measured .
this is important as any effect within the supernova which could drive the temperatures together without new physics , for example that of ref .
@xcite , or an effect which suppresses the mixing of the neutrino flavours in transit to earth would weaken our limits , especially in the case of @xmath52 detection .
hence it is possible that the limit from coherent nuclear scattering is more robust to such effects , beyond the simplistic model of the star during the supernova event which we have used here .
indeed this motivates the construction of such detectors up to 100 tonnes or more in mass , as an effective way of detecting all flavours of supernova neutrinos with high precision .
we now consider such @xmath0 particles within a wider context by comparing our projected limits with other constraints .
these particles should be only weakly coupled at most to visible matter to avoid having been already detected , however even so there are other methods of probing their interactions .
indeed the temperature of the neutrinos is not the only method of using a supernova to constrain new particles , which interact only weakly with visible matter .
for example the @xmath0 particle needs to have a mean free path for scattering which is shorter than the size of the hot star during the supernova event , to avoid excessively rapid cooling @xcite i.e. a new particle produced thermally but which scatters less often with the stellar matter than neutrinos do could make the energy loss rate too rapid .
hence the @xmath0 particle likely needs interactions in addition to those with neutrinos in order to evade this cooling bound , and so our assumption of thermal equilibrium for @xmath0 is reasonable . to study thermal equilibrium in more detail we follow a similar argument using equation ( [ eqn : diff ] ) .
broadly speaking if @xmath8 is small then @xmath5 is close to @xmath10 and so the particles are in thermal equilibrium ( or close to it ) . in this case
a small value of @xmath8 means that the @xmath0 particles scatter frequently with the stellar matter . indeed assuming that @xmath0 maintains its equilibrium through interactions with the baryonic matter of the star we can make use of the mean free path for scattering @xmath87 to estimate the required scattering cross section .
we need that @xmath88 , where @xmath89 is the linear size of the hot star . assuming that @xmath90 , where @xmath91 is the scattering cross section between baryonic matter and @xmath0 we have that @xmath92 .
the value of @xmath17 depends on the model for the hot star i.e. equations ( [ eqn : nbprof ] ) and ( [ eqn : tprof ] ) and the temperature .
if we assume that we need the @xmath0 to be in thermal equilibrium at temperatures above 1 mev then using parameters from model a we have that @xmath93 @xmath29 . hence if @xmath94 km we have the approximate requirement that @xmath95 @xmath2 for thermal equilibrium of the @xmath0 particles down to temperatures of 1 mev
however this bound could be several orders of magnitude higher for more extreme models of the hot star during neutrino thermal production .
equation ( [ eqn : diff ] ) also tells us that the self - annihilation cross section of @xmath0 should be non - zero for thermal equilibrium to be established , though this depends partly on whether @xmath0 can decay . assuming that @xmath0 has a lifetime much longer than the time - period for thermal production then the larger the value of @xmath8 the larger the self - annihilation cross section @xmath9 needs to be for @xmath5 to be equal ( or close ) to @xmath10 . solving equation ( [ eqn : diff ] ) numerically for the number density @xmath5 of the @xmath0 particles with @xmath96
@xmath2 confirms that @xmath0 are in thermal equilibrium down to temperatures of around 1 mev for a self - annihilation cross section @xmath97 @xmath98s@xmath99 , which is the scale required for @xmath0 to be the dark matter of the universe .
values of @xmath9 larger than this are also allowed for thermal equilibrium , however in this case the relic density of @xmath0 will be too small for it to be dark matter . bounds also exist on the temperature of @xmath100 and @xmath101 produced in supernovae from the abundances of light elements @xcite .
this constraint is complimentary to the one studied in this work and could in principle be used to constrain @xmath42 .
as mentioned previously thermally produced new states will disrupt nucleosynthesis unless their mass is above @xmath102 mev @xcite and there are bounds from the number of relativistic degrees of freedom in the early universe @xcite
. there also exist constraints on the scattering of @xmath0 with neutrinos from the damping of structure in the universe @xcite , though this only applies if @xmath0 forms the dominant component of dark matter ( set by the size of @xmath9 ) . at @xmath3 confidence for a 100 kilo - tonne experiment looking for neutrinos through inverse beta decay ( ibd ) capture , with the upper limit from requiring that dark matter scattering with neutrinos does not destroy small - scale structure in the universe @xcite .
the latter only applies if @xmath0 forms the dominant component of the dark matter.,scaledwidth=49.0% ] in figure [ fig : lims_mass ] we show our projected limit for an experiment with a mass of 100 kilo - tonnes detecting @xmath52 via inverse beta decay for a range of @xmath0 masses above the bound from nucleosynthesis @xcite . our projected limit is stronger than that from the damping of structure in the universe @xcite for masses up to around 100 mev , above which the density of @xmath0 in the hot star becomes too small due to boltzmann suppression to lead to any significant scattering with neutrinos , unless @xmath42 is very large .
if a supernova event were to occur in our own galaxy then a large number of neutrino events should be detected on earth in the space of a few seconds .
the majority of the neutrinos are produced thermally with a temperature depending on their flavour : electron neutrinos scatter the most with baryonic matter and so have the smallest temperature , while electron anti - neutrinos are slightly hotter and the other heavy flavours hotter still @xcite . in this work we have considered the prospect of additional mev - mass ` dark ' particles
@xmath0 being produced thermally during the supernova , which scatter with neutrinos @xcite .
we have focused on the effect these new states would have on the temperatures of the different neutrino flavours due to scattering , with cross section @xmath42 .
this leads to an extra term in the diffusion coefficient @xmath8 i.e. equation ( [ eqn : diff_chi ] ) , which alters the hierarchy of the temperatures for each flavour , as shown in figure [ fig : temps_plot ] . however quantifying this effect
is made difficult by the uncertainty in the modelling of the density and temperature of the star during thermal production .
our main emphasis has been on predicting the ability of current and future experiments to be able to measure the neutrino spectrum with enough precision to place constraints on @xmath42 .
we show some example spectra in figure [ fig : spectra_sigmas ] .
since the temperatures of the different neutrino flavours depend sensitively on the model for the star during thermal production of neutrinos ( and potentially @xmath0 also ) , we have performed a bayesian analysis where we marginalise over the parameters of a simple analytic stellar model ( equations ( [ eqn : nbprof ] ) and ( [ eqn : tprof ] ) ) .
we have considered experiments looking at coherent nuclear scattering , such as direct detection experiments , and those looking for @xmath52 via inverse beta decay capture , such as super and hyper kamiokande @xcite .
as can be seen in figure [ fig : lims ] , we have found that super kamiokande could place an upper limit at the level of @xmath1 @xmath2 for 10 mev mass @xmath0 particles and hyper kamiokande could be able to set a limit an order of magnitude stronger still .
furthermore as shown in figure [ fig : lims_mass ] our projected limits are potentially the strongest constraint on @xmath0 scattering with neutrinos up to a mass around 100 mev .
however these projections are essentially the ` best case scenario ' in the sense that we assume there are no additional effects on top of our simplified model for the star during neutrino production which could bring the temperatures of @xmath52 and @xmath103 closer together @xcite .
such effects could weaken these limits considerably .
indeed our analysis was performed using a simple empirical parametrisation to model the star during neutrino production , in order to understand how the uncertainties in this model impact our ability to constrain @xmath42 .
it is likely that there are further effects which will alter the expected temperatures of the neutrino species @xcite .
however we expect that the value of @xmath42 where the temperatures of the different flavours become equal will remain fairly robust to additional modelling uncertainties
. for detectors looking at coherent nuclear scattering the limits should in principle be more robust to such effects , since they will be able to measure all neutrino flavours through neutral - current scattering .
however experiments which exploit this channel , for example dark matter direct detection experiments @xcite , are generally much less massive than super kamiokande and so can not exploit as many statistics to measure the neutrino spectrum .
this motivates construction of an 100 tonne mass liquid - xenon experiment or similar as potentially the ultimate neutral - current supernova detector , with regards to effects from @xmath0-@xmath104 scattering .
the author thanks john beacom and malcolm fairbairn for helpful comments and suggestions .
the research leading to these results has received funding from the european research council through the project darkhorizons under the european union s horizon 2020 program ( erc grant agreement no.648680 ) .
30ifxundefined [ 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 10.1103/physrevlett.58.1490 [ * * , ( ) ] link:\doibase 10.1103/physrevd.76.083007 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.68.3834 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevd.51.1499 [ * * , ( ) ] , link:\doibase 10.1086/375130 [ * * , ( ) ] , link:\doibase 10.1093/ptep / ptv079 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.54.1194 [ * * , ( ) ] link:\doibase 10.1103/physrevd.83.113006 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.96.211302 [ * * , ( ) ] , @noop ( ) , link:\doibase 10.1086/323379 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.104.251101 [ * * , ( ) ] \doibase 10.1051/0004 - 6361:20053783 [ * * , ( ) ] , link:\doibase 10.1393/ncr / i2016 - 10120 - 8 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.74.043517 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.77.043516 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.70.043526 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/s0168-9002(03)00425-x [ * * , ( ) ] @noop ( ) , link:\doibase 10.1088/1742 - 6596/375/1/012028 [ * * , ( ) ] , link:\doibase 10.1088/1475 - 7516/2014/01/044 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.89.023524 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.76.033007 [ * * , ( ) ] , link:\doibase 10.1088/1475 - 7516/2011/10/019 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.77.113007 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.103.051105 [ * * , ( ) ] , link:\doibase 10.1088/1475 - 7516/2013/08/041 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.94.231101 [ * * , ( ) ] link:\doibase 10.1088/1475 - 7516/2014/05/011 [ * * , ( ) ] , | a supernova event in our own galaxy will result in a large number of neutrinos detected on earth within the time - frame of a few seconds .
these neutrinos will have been produced thermally with , in principle , three distinct temperatures for the electron , anti - electron and remaining heavy flavours respectively .
we revisit the possibility that new mev - mass particles @xmath0 are also produced thermally during the event , which scatter with the neutrinos and alter their temperatures .
our main emphasis is on the detectability of this effect using the neutrino spectrum , given the large uncertainty on the temperature and density profiles of the stellar matter . by marginalising over the parameters of a simple analytic model for the stellar profile , we find that super kamiokande could place an upper limit on the scattering cross section at the level of @xmath1 @xmath2 for 10 mev mass @xmath0 particles at @xmath3 confidence .
a direct - detection - like experiment would be less susceptible to systematic uncertainties in the neutrino production and mixing , but this would need a target mass around 100 tonnes in order to acquire enough statistics to compete with super kamiokande . |
a key prediction of a number of simple single - field slow - roll inflationary models is that they can not generate detectable non - gaussianity of the cosmic microwave background ( cmb ) temperature fluctuations within the level of accuracy of the wilkinson microwave anisotropy probe ( wmap ) @xcite .
there are , however , several inflationary models that can generate non - gaussianity at a level detectable by the wmap .
these non - gaussian scenarios comprise models based upon a wide range of mechanisms , including special features of the inflation potential and violation of one of the following four conditions : single field , slow roll , canonical kinetic energy , and initial bunch - davies vacuum state .
thus , although convincing detection of a fairly large primordial non - gaussianity in the cmb data would not rule out all inflationary models , it would exclude the entire class of stationary models that satisfy _ simultaneously _ these four conditions ( see , e.g. , refs . @xcite ) . moreover
, a null detection of deviation from gaussianity would rule out alternative models of the early universe ( see , for example , refs .
thus , a detection or nondetection of primordial non - gaussianity in the cmb data is crucial not only to discriminate ( or even exclude classes of ) inflationary models but also to test alternative scenarios , offering therefore a window into the physics of the primordial universe .
however , there are various non - primordial effects that can also produce non - gaussianity such as , e.g. , unsubtracted foreground contamination , unconsidered point sources emission and systematic errors @xcite .
thus , the extraction of a possible primordial non - gaussianity is not a simple endeavor . in view of this , a great deal of effort has recently gone into verifying the existence of non - gaussianity by employing several statistical estimators @xcite ( for related articles see , e.g. , refs .
different indicators can in principle provide information about multiple forms of non - gaussianity that may be present in wmap data .
it is therefore important to test cmb data for deviations from gaussianity by using a range of different statistical tools to quantify or constrain the amount of any non - gaussian signals in the data , and extract information on their possible origins .
a number of recent analyses of cmb data performed with different statistical tools have provided indications of either consistency or deviation from gaussianity in the cmb temperature fluctuations ( see , e.g. , ref .
@xcite ) . in a recent paper
@xcite we proposed two new large - angle non - gaussianity indicators , based on skewness and kurtosis of large - angle patches of cmb maps , which provide measures of the departure from gaussianity on large angular scales .
we used these indicators to search for the large - angle deviation from gaussianity in the three and five - year single frequency maps with a _
kq75 _ mask , and found that while the deviation for the q , v , and w masked maps are within the @xmath0 expected values of monte - carlo ( mc ) statistically gaussian cmb maps , there is a strong indication of deviation from gaussianity ( @xmath1 off the mc ) in the k and ka masked maps .
most of the gaussianity analyses with wmap data have been carried out by using cmb temperature fluctuation maps ( raw and clean ) in the frequency bands q , v and w or some combination of these maps . in these analyses , in order to deal with the diffuse galactic foreground emission , masks such as , for example , _
kq75 _ and _ kp0 _ have been used .
however , sky cuts themselves can potentially induce bias in gaussianity analyses , and on the other hand full - sky maps seem more appropriate to test for gaussianity in the cmb data .
thus , a pertinent question that arises is how the analysis of gaussianity made in ref .
@xcite is modified if whole - sky foreground - reduced cmb maps are used .
our primary objective in this paper is to address this question by extending the analysis of ref .
@xcite in three different ways .
first , we use the same statistical indicators to carry out a new analysis of gaussianity of the available _ full - sky foreground - reduced _ five - year and seven - year cmb maps @xcite .
second , since in these maps the foreground is reduced through different procedures each of the resulting maps should be tested for gaussianity .
thus , we make a quantitative analysis of the effects of distinct cleaning processes in the deviation from gaussianity , quantifying the level of non - gaussianity for each foreground reduction method .
third , we study quantitatively the consequences for the gaussianity analysis of masking the foreground - reduced maps with the _ kq75 _ mask .
an interesting outcome is that this mask lowers significantly the level of deviation from gaussianity even in the foreground - reduced maps , rendering therefore information about the suitability of the foreground - reduced maps as gaussian reconstructions of the full - sky cmb .
the chief idea behind our construction of the non - gaussianity indicators is that a simple way of accessing the deviation from gaussianity distribution of the cmb temperature fluctuations is by calculating the skewness @xmath2 , and the kurtosis @xmath3 from the fluctuations data , where @xmath4 and @xmath5 are the third and fourth central moments of the distribution , and @xmath6 is its variance . clearly calculating @xmath7 and @xmath8 from the whole sky temperature fluctuations data
would simply yield two dimensionless numbers , which are rough measures of deviation from gaussianity of the temperature fluctuation distribution .
however , one can go further and obtain a great number of values associated to directional information of deviation from gaussianity if instead one takes a discrete set of points @xmath9 homogeneously distributed on the celestial sphere @xmath10 as the center of spherical caps of a given aperture @xmath11 and calculate @xmath12 and @xmath13 from the cmb temperature fluctuations of each spherical cap .
the values @xmath12 and @xmath13 can then be taken as measures of the non - gaussianity in the direction @xmath14 of the center of the spherical cap @xmath15 .
such calculations for the individual caps thus provide quantitative information ( @xmath16 values ) about possible violation of gaussianity in the cmb data .
this procedure is a constructive way of defining two discrete functions @xmath7 and @xmath8 ( defined on @xmath17 from the temperature fluctuations data , and can be formalized through the following steps ( for more details , see ref .
@xcite ) : 1 . take a discrete set of points @xmath18 homogeneously distributed on the cmb celestial sphere @xmath10 as the centers of spherical caps of a given aperture @xmath11 ; 2 .
calculate for each spherical cap @xmath19 the skewness ( @xmath12 ) and kurtosis ( @xmath13 ) given , respectively , by @xmath20 where @xmath21 is the number of pixels in the @xmath22 cap , @xmath23 is the temperature at the @xmath24 pixel , @xmath25 is the cmb mean temperature in the @xmath22 cap , and @xmath6 is the standard deviation . clearly , the values @xmath12 and @xmath13 obtained in this way for each cap can be viewed as a measure of non - gaussianity in the direction of the center of the cap @xmath14 ; 3 .
patching together the @xmath12 and @xmath13 values for each spherical cap , one obtains our indicators , i.e. , discrete functions @xmath26 and @xmath27 defined over the celestial sphere , which can be used to measure the deviation from gaussianity as a function of the angular coordinates @xmath28 .
the mollweid projection of skewness and kurtosis functions @xmath26 and @xmath27 are nothing but skewness and kurtosis maps , hereafter we shall refer to them as @xmath29map and @xmath30map , respectively .
now , since @xmath26 and @xmath27 are functions defined on @xmath10 they can be expanded into their spherical harmonics in order to have their power spectra @xmath31 and @xmath32 .
thus , for example , for the skewness indicator @xmath26 one has @xmath33 and can calculate the corresponding angular power spectrum @xmath34 which can be used to quantify the angular scale of the deviation from gaussianity , and also to calculate the statistical significance of such deviation .
obviously , similar expressions hold for the kurtosis @xmath27 . in the next section
we shall use the statistical indicators @xmath26 and @xmath27 to test for gaussianity the available foreground - reduced maps obtained from the five - year wmap data .
the wmap team has released high angular resolution five - year maps of the cmb temperature fluctuations in the five frequency bands k ( @xmath35 ghz ) , ka ( @xmath36 ghz ) , q ( @xmath37 ghz ) , v ( @xmath38 ghz ) , and w ( @xmath39 ghz ) .
they have also produced a full - sky foreground - reduced internal linear combination ( ilc ) map which is formed from a weighted linear combination of these five frequency band maps in which the weights are chosen in order to minimize the galactic foreground contribution .
it is well known that the _ first - year _ ilc map is inappropriate for cmb scientific studies @xcite .
however , in the _ five - year _ ( also in the three - year and seven - year ) version of this map a bias correction has been implemented as part of the foreground cleaning process , and the wmap team suggested that this map is suitable for use in large angular scales ( low @xmath40 ) analyses although they admittedly have not performed non - gaussian tests on this version of the ilc map @xcite . notwithstanding
the many merits of the five - year ilc procedure , some cleaning features of this ilc approach have been considered , and two variants have been proposed recently . in the first approach
the frequency dependent weights were determined in harmonic space @xcite , while in the second the foreground is reduced by using needlets as the basis of the cleaning process @xcite .
thus , two new full - sky foreground - cleaned maps have been produced with the wmap five - year data , namely the harmonic ilc ( hilc ) @xcite and the needlet ilc ( nilc ) ( for more details see refs .
@xcite ) . in the next section ,
we use the full - sky foreground - reduced ilc , hilc and nilc maps with the same smoothed @xmath41 resolution ( which is the resolution of the ilc map ) as the input maps from which we calculate the @xmath26 and @xmath27 maps , and then we compute the associated power spectra in order to carry out a statistical analysis to quantify the levels of deviation from gaussianity.jkim/hilc/ and http://www.apc.univ-paris7.fr/apc_cs/recherche/adamis/cmb_wmap-en.php . ] in order to minimize the statistical noise , in the calculations of skewness and kurtosis maps ( @xmath29map and @xmath30map ) from the foreground - reduced maps , we have scanned the celestial sphere with spherical caps of aperture @xmath42 , centered at @xmath43 points homogeneously generated on the two - sphere by using the healpix code @xcite . in other words ,
the point - centers of the spherical caps are the center of the pixels of a homogeneous pixelization of the @xmath10 generated by healpix with @xmath44 .
we emphasize , however , that this pixelization is only a practical way of choosing the centers of the caps homogeneously distributed on @xmath10 .
it is not related to the pixelization of the above - mentioned ilc , hilc and nilc input maps that we have utilized to calculate both the @xmath7 and @xmath8 maps from which we compute the associated power spectra .
figures [ fig1 ] and [ fig2 ] show examples of @xmath7 and @xmath8 maps obtained from the foreground - reduced nilc full - sky and _ kq75 _ maps .
the panels of these figures clearly show regions with higher and lower values ( hot and cold spots ) of @xmath45 and @xmath46 , which suggest _ large - angle _ multipole components of non - gaussianity .
we have also calculated similar maps ( with and without the _ kq75 _ mask ) from the ilc and hilc maps . however , since these maps provide only _ qualitative _ information , to avoid repetition we only depict the maps of figs . [ fig1 ] and [ fig2 ] merely for illustrative purpose . in order to obtain _ quantitative _ information about the large angular scale ( low @xmath40 ) distributions for the non - gaussianity @xmath7 and @xmath8 maps obtained from the available full - sky foreground - reduced five - year maps , we have calculated the ( low @xmath40 ) power spectra @xmath31 and @xmath32 for these maps .
the statistical significance of these power spectra is estimated by comparing with the corresponding multipole values of the averaged power spectra @xmath47 and @xmath48 calculated from maps obtained by averaging over @xmath49 monte - carlo - generated statistically gaussian cmb maps.cdm model , obtained by randomizing the temperature components @xmath50 within the cosmic variance limits . ] throughout the paper the mean quantities are denoted by overline . before proceeding to a statistical analysis ,
let us describe with some detail our calculations .
for the sake of brevity , we focus on the skewness indicator @xmath7 , but a completely similar procedure was used for the kurtosis indicator @xmath8 .
we generated @xmath49 mc gaussian ( _ scrambled _ ) cmb maps , which are then used to generate @xmath49 skewness @xmath29maps , from which we calculate @xmath49 power spectra : @xmath51 ( @xmath52 is an enumeration index , and @xmath53 ) . in this way , for each fixed multipole component @xmath54 we have @xmath55 multipole values from which we calculate the mean value @xmath56 . from this mc process
we have at the end ten mean multipole values @xmath47 , each of which are then used for a comparison with the corresponding multipole values @xmath31 ( obtained from the input map ) in order to evaluate the statistical significance of the multipole components @xmath31 . to make this comparison easier , instead of using the angular power spectra @xmath31 and
@xmath32 themselves , we employed the _ differential _ power spectra @xmath57 and @xmath58 , which measure the deviation of the skewness and kurtosis multipole values ( calculated from the foreground - reduced maps ) from the mean multipoles @xmath47 and @xmath48 ( calculated from the gaussian maps ) .
thus , for example , to study the statistical significance of the quadrupole component of the skewness from hilc map @xmath59 ( say ) we calculate the deviation @xmath60 , where the mean quadrupole value @xmath61 is calculated from the @xmath52 quadrupole values of the mc gaussian maps .
it is interesting to note that the deviation from gaussianity as measured by our indicators is greater for the ilc7 than for the ilc5 input map .
concerning this point some words of clarification are in order here .
first , we note that the details of the algorithm used to compute the ilc7 maps are the same as those of the ilc5 map . however , to take into account the most recent updates to the calibration and beams , the frequency weights for each of the 12 regions ( in which the sky is subdivided in the ilc method ) are slightly different in the calculation of the ilc7 map .
second , the difference between the ilc7 and ilc5 maps is a map whose small - scale differences are consistent with the pixel noise , but with a large - scale dipolar component , with the large - scale differences being consistent with a change in dipole of 6.7 @xmath62k@xcite .
thus , the resultant ilc7 map is not indistinguishable from the ilc5 map , and the differences between them have been captured by our indicators .
figure [ fig6 ] shows the differential power spectra calculated from a five - year and seven year version of the foreground - reduced ilc maps with a _
kq75 _ mask .
this figure along with fig .
[ fig5 ] show a significant reduction in the level of deviation from gaussianity when both ilc5 and ilc7 are masked . to quantify this reduction
we have calculated @xmath63 for these input maps with the _ kq75 _ mask , and have collected the results in table [ chi2-table - ilc5 - 7_kq75 ] .
the comparison of table [ chi2-table - ilc5 - 7_full - sky ] and table [ chi2-table - ilc5 - 7_kq75 ] shows quantitatively the reduction of the level of gaussianity for the case of cmb masked maps
. 999 v. acquaviva , n. bartolo , s. matarrese , and a. riotto , nucl . phys .
b * 667 * , 119 ( 2003 ) ; j. maldacena , jhep * 0305 * 013 ( 2003 ) ; m. liguori , f.k .
hansen , e. komatsu , s. matarrese , and a. riotto , * 73 * , 043505 ( 2006 ) .
n. bartolo , e. komatsu , s. matarrese , and a. riotto , phys . rept . * 402 * , 103 ( 2004 ) .
e. komatsu _ et al .
_ , arxiv:0902.4759v4 [ astro-ph.co ] . b.a .
bassett , s. tsujikawa , and d. wands , rev .
78 * , 537 ( 2006 ) ; a. linde , lect .
notes phys .
* 738 * , 1 ( 2008 ) .
k. koyama , s. mizuno , f. vernizzi , and d. wands , _ jcap _ 11 ( 2007 ) 024 ; e.i .
buchbinder , j. khoury and b.a .
ovrut , * 100 * , 171302 ( 2008 ) ; j .-
lehners and p.j .
steinhardt , * 77 * , ( 2008 ) 063533 ; y .- f .
cai , w. xue , r. brandenberger , and x. zhang , _ jcap _ 05 ( 2009 ) 011 ; y .- f .
cai , w. xue , r. brandenberger , and x. zhang , _ jcap _ 06 ( 2009 ) 037 .
chiang , p.d .
naselsky , o.v .
verkhodanov , and m.j .
way , * 590 * , l65 ( 2003 ) .
p. cabella , d. pietrobon , m. veneziani , a. balbi , r. crittenden , g. de gasperis , c. quercellini , and n. vittorio , arxiv:0910.4362 .
e. komatsu et al . , astrophys .
* 148 * , 119 ( 2003 ) ; d.n .
spergel et al . , astrophys .
* 170 * , 377 ( 2007 ) ; e. komatsu et al . ,
* 180 * , 330 ( 2009 ) ; p. vielva , e. martnez - gonzlez , r.b .
barreiro , j.l .
sanz , and l. cayn , * 609 * , 22 ( 2004 ) ; m. cruz , e. martnez - gonzlez , p. vielva , and l. cayn , mon . not .
. soc . * 356 * , 29 ( 2005 ) ; m. cruz , l. cayn , e. martnez - gonzlez , p. vielva , and j. jin , * 655 * , 11 ( 2007 ) ; l. cayn , j. jin , and a. treaster , mon . not .
. soc . * 362 * , 826 ( 2005 ) ; lung - y chiang , p.d .
naselsky , int . j. mod .
d * 15 * , 1283 ( 2006 ) ; j.d .
mcewen , m.p .
hobson , a.n .
lasenby , and d.j .
mortlock , mon . not .
r. astron .
soc . * 371 * , l50 ( 2006 ) ; j.d .
mcewen , m.p .
hobson , a.n .
lasenby , and d.j .
mortlock , mon . not .
. soc . * 388 * , 659 ( 2008 ) ; a. bernui , c. tsallis , and t. villela , europhys .
lett . * 78 * , 19001 ( 2007 ) ; l .- y .
chiang , p.d .
naselsky , and p. coles , * 664 * , 8 ( 2007 ) ; c .- g .
park , mon . not .
r. astron . soc . *
349 * , 313 ( 2004 ) ; h.k .
eriksen , d.i .
novikov , p.b .
lilje , a.j .
banday , and k.m .
grski , * 612 * , 64 ( 2004 ) ; m. cruz , m. tucci , e. martnez - gonzlez , and p. vielva , mon . not .
r. astron .
369 * , 57 ( 2006 ) ; m. cruz , n. turok , p. vielva , e. martnez - gonzlez , and m. hobson , science * 318 * , 1612 ( 2007 ) ; p. mukherjee and y. wang , * 613 * , 51 ( 2004 ) ; d. pietrobon , p. cabella , a. balbi , g. de gasperis , and n. vittorio , mon . not .
. soc . * 396 * , 1682 ( 2009 ) ; d. pietrobon , p. cabella , a. balbi , r. crittenden , g. de gasperis , and n. vittorio , mon . not .
. soc . * 402 * , l34 ( 2010 ) ; y. ayaita , m. weber , and c. wetterich , * 81 * , 023507 ( 2010 ) ; p. vielva and j.l .
sanz , mon . not .
astron . soc . *
397 * , 837 ( 2009 ) ; b. lew , _
jcap _ 08 ( 2008 ) 017 ; a. bernui and m.j .
rebouas , int . j. mod
a * 24 * , 1664 ( 2009 ) ; m. kawasaki , k. nakayama , t. sekiguchi , t. suyama , and f. takahashi , _ jcap _ 11 ( 2008 ) 019 ; m. kawasaki , k. nakayama , and f. takahashi , _ jcap _ 01 ( 2009 ) 026 ; m. kawasaki , k. nakayama , t. sekiguchi , t. suyama , and f. takahashi , _ jcap _ 01 ( 2009 ) 042 ; m. cruz , e. martnez - gonzlez , and p. vielva , arxiv:0901.1986 [ astro - ph ] .
e. martnez - gonzlez , arxiv:0805.4157 [ astro - ph ] ; y. wiaux , p. vielva , e. martnez - gonzlez , and p. vandergheynst , * 96 * , 151303 ( 2006 ) ; l.r .
abramo , a. bernui , i.s .
ferreira , t. villela , and c.a .
wuensche , * 74 * , 063506 ( 2006 ) ; p. vielva , y. wiaux , e. martnez - gonzlez , and p. vandergheynst , new astron .
* 50 * , 880 ( 2006 ) ; p. vielva , y. wiaux , e. martnez - gonzlez , and p. vandergheynst , mon . not .
r. astron .
soc . , * 381 * , 932 ( 2007 )
copi , d. huterer , d.j .
schwarz , and g.d .
starkman , * 75 * , 023507 ( 2007 ) ; c.j .
copi , d. huterer , d.j .
schwarz , and g.d .
starkman , mon . not .
. soc . * 399 * , 295 ( 2009 ) ; k. land and j. magueijo , mon . not .
. soc . * 378 * , 153 ( 2007 ) ; b. lew , _ jcap _ 09 ( 2008 ) 023 ; a. bernui , * 78 * , 063531 ( 2008 ) ; a. bernui , * 80 * , 123010 ( 2009 ) ; p.k .
samal , r. saha , p. jain , and j.p .
ralston , mon . not .
. soc . * 385 * , 1718 ( 2008 ) ; p.k .
samal , r. saha , p. jain , and j.p .
ralston , mon . not .
. soc . * 396 * , 511 ( 2009 ) ; a. bernui , b. mota , m.j .
rebouas , and r. tavakol , astron .
& astrophys .
* 464 * , 479 ( 2007 ) ; a. bernui , b. mota , m.j .
rebouas , and r. tavakol , int .
d * 16 * , 411 ( 2007 ) ; t. kahniashvili , g. lavrelashvili , and b. ratra , * 78 * , 063012 ( 2008 ) ; l.r .
abramo , a. bernui , and t.s .
pereira , _ jcap _ 12 ( 2009 ) 013 ; f.k .
hansen , a.j .
banday , k.m .
grski , h.k .
eriksen , and p.b .
lilje , * 704 * , 1448 ( 2009 ) . | a detection or nondetection of primordial non - gaussianity by using the cosmic microwave background radiation ( cmb ) data is crucial not only to discriminate inflationary models but also to test alternative scenarios .
non - gaussianity offers , therefore , a powerful probe of the physics of the primordial universe .
the extraction of primordial non - gaussianity is a difficult enterprise since several effects of non - primordial nature can produce non - gaussianity .
given the far - reaching consequences of such a non - gaussianity for our understanding of the physics of the early universe , it is important to employ a range of different statistical tools to quantify and/or constrain its amount in order to have information that may be helpful for identifying its causes .
moreover , different indicators can in principle provide information about distinct forms of non - gaussianity that can be present in cmb data .
most of the gaussianity analyses of cmb data have been performed by using part - sky frequency , where the masks are used to deal with the galactic diffuse foreground emission .
however , full - sky map seems to be potentially more appropriate to test for gaussianity of the cmb data . on the other hand
, masks can induce bias in some non - gaussianity analyses .
here we use two recent large - angle non - gaussianity indicators , based on skewness and kurtosis of large - angle patches of cmb maps , to examine the question of non - gaussianity in the available full - sky five - year and seven - year wilkinson microwave anisotropy probe ( wmap ) maps .
we show that these full - sky foreground - reduced maps present a significant deviation from gaussianity of different levels , which vary with the foreground - reducing procedures .
we also make a gaussianity analysis of the foreground - reduced five - year and seven - year wmap maps with a _
kq75 _ mask , and compare with the similar analysis performed with the corresponding full - sky foreground - reduced maps .
this comparison shows a significant reduction in the levels of non - gaussianity when the mask is employed , which provides indications on the suitability of the foreground - reduced maps as gaussian reconstructions of the full - sky cmb . |
a point defect in a crystal is an entity that causes an interruption in the lattice periodicity . in this paper ,
the terminology and definitions in agreement with m. lannoo and j. bourgoin @xcite are used in relation to defects .
vacancies and interstitials are produced in materials exposed to radiation in equal quantities ; they are the primary radiation defects , being produced either by the incoming particle , or as a consequence of the subsequent collisions produced by the primary recoil in previous papers @xcite we calculated the concentration of radiation induced primary defects ( cpd ) in semiconductors exposed to hadron irradiation . due to the important weight of annealing processes , as well as to their very short time scale , cpd is not a measurable physical quantity . in silicon ,
vacancies and interstitials are essentially unstable and interact via migration , recombination , annihilation or produce other defects .
the system evolves toward natural equilibrium .
the problem of the annealing of radiation induced defects in semiconductor materials is not new .
several models , empirical or more theoretic , have been previously proposed to explain these phenomena .
damask and dienes developed , in the 1960s , a theoretical model that considers , as distinct cases , the kinetics of vacancy - interstitial annihilation with interstitial trapping at impurities @xcite , of vacancy - interstitial annihilation with interstitial trapping at impurities @xcite , and of vacancy - interstitial annihilation with diinterstitial formation @xcite . in the first two cases , analytical solutions have been obtained by the authors , while in the last case only numerical solutions are possible .
the escape of interstitials and vacancies from clusters and their subsequent interactions have been modelled by tsveybak et al .
phenomenological approaches of the annealing could be found in moll et al .
@xcite and bates and co - workers @xcite and references cited therein .
most of the old studies were dedicated to electronic devices , made from low resistivity ( in the @xmath0 range ) silicon .
recently , the behaviour of the silicon material with medium and high resistivity ( tens @xmath0 - @xmath1 ) in intense hadron fields started to present very much interest , these materials representing a major option for the detectors at the new generation of particle colliders and for space applications . in the present paper , the time evolution of the concentration of primary defects induced by pion irradiation in silicon
is studied in the frame of a simplified phenomenological model based on direct interactions between the primary induced defects and the impurities present in the material .
the starting hypothesis of the physical model is that the irradiation process is unique at the determined particle fluence , and that the time of irradiation is shorter that all characteristic times of annealing . in the frame of the model , equal concentrations of vacancies and interstitials
are supposed to be produced by irradiation , in much greater concentrations than the corresponding thermal equilibrium values , characteristic to each temperature .
both the pre - existing defects and those produced by irradiation , as well as the impurities , are assumed to be randomly distributed in the solid .
an important part of the vacancies and interstitials annihilate .
the sample contains certain concentrations of impurities which can trap interstitials and vacancies respectively , and form stable defects . in the present paper , vacancy - interstitial annihilation , interstitial migration to sinks ,
vacancy and interstitial impurity complex formation as well as divacancy formation are considered .
the sample could contain more impurities that trap vacancies or interstitials , and in this case all processes are to be taken into account .
the following notations are used : @xmath2 - monovacancy concentration ; @xmath3 - free interstitial concentration , @xmath4 - total impurity `` 1 '' concentration ( impurity `` 1 '' traps interstitials and forms the complex @xmath5 ) ; @xmath5 - interstitial - impurity concentration : one interstitial trapped for one complex formed ; @xmath6 - total impurity `` 2 '' concentration ( impurity `` 2 '' traps vacancies and forms the complex @xmath7 ) ; @xmath7 - vacancy - impurity concentration : one vacancy trapped for one complex formed ; @xmath8 - divacancy concentration .
all concentrations are expressed as atomic fractions .
the chemical scheme for reaction rates has been used , and the differential equations corresponding to the above picture are : @xmath9 @xmath10 @xmath11 @xmath12 @xmath13 the bimolecular recombination law of interstitials and vacancies is supposed to be a valid approximation for the present discussion , because at the concentrations of vacancies of interest , only a small fraction of defects anneal by correlated annihilation if their distribution is random ( see the discussion in @xcite ) .
if @xmath14 is the total defect concentration , expressed as atomic fraction : @xmath15 then it satisfies the differential equation : @xmath16 the initial conditions , at @xmath17 , are : at the end of the irradiation , there are equal concentrations of interstitials and vacancies : @xmath18 ; the concentrations of impurities are @xmath19 and @xmath20 respectively , there are no complexes in the sample : @xmath21 , and no divacancies @xmath22 .
the reaction constants @xmath23 and @xmath24 are determined by the diffusion coefficient for the interstitial atom to a substitutional trap , and therefore @xmath25 : @xmath26 where @xmath27 is the activation energy of interstitial migration and @xmath28 the vibrational frequency .
the reaction constant @xmath29 ( related to interstitial migration to sinks ) is proportional to the sink concentration @xmath30 : @xmath31 with @xmath32 the jump distance .
@xmath33 characterises the decomposition of a complex into an impurity and an interstitial , and is given by : @xmath34 @xmath35 and @xmath36 describe the formation and decomposition of vacancy - impurity complexes , while @xmath37 and @xmath38 the formation and decomposition of divacancies . and
are given by : @xmath39 with @xmath40 the activation energy for vacancy migration ; @xmath41 @xmath42 @xmath43 where @xmath44 is the binding energy of @xmath5 , @xmath45 the binding energy of @xmath7 and @xmath46 is the corresponding binding energy of divacancies . due to the mathematical difficulties to solve analytically the complete differential equation system , some simplifications are necessary .
the interstitials are much more mobile in silicon in respect to vacancies , and are characterised by an activation energy of migration a factor of two times smaller .
this difference between the activation energies @xmath27 and @xmath40 respectively justifies the introduction of two time scales , and the separate study of the processes involving interstitials and vacancies respectively . in the first stage , vacancy interstitial annihilation and interstitials migration to sinks
are studied .
the concentration of interstitials decays much rapidly than the concentration of vacancies .
a cut - off condition for @xmath3 is imposed ( a @xmath47 times decrease of interstitial concentration , @xmath47 being an adjustable parameter of the model ) .
the vacancy concentration determined by this procedure is the initial value for the processes comprised in the second stage , and will be denoted by @xmath48 .
the formation of complexes by vacancies is considered less important , and is neglected in the following discussion .
so , for the `` first stage '' , after some simple manipulations , from ( 1 ) and ( 2 ) , the equation : @xmath49 has been obtained , with the solution : @xmath50 and : @xmath51 } { k_2\ln \left ( v\left ( t\right ) \right ) } \ ] ] imposing the cut - off condition for the concentration of interstitials , both @xmath48 and the characteristic time could be found .
as specified , @xmath48 is used as initial vacancy concentration for the second step in the analysis , where the equations ( 4 ) and ( 5 ) are considered .
@xmath48 depends on @xmath52 and @xmath47 , and it is temperature independent .
this system of equations , expressing the kinetics of divacancy and vacancy - impurity formation , have no analytical solution .
these processes are governed by the initial concentrations of vacancies and impurities .
if the impurity that traps vacancies is phosphorus , the limiting cases correspond to low initial doping concentration ( high resistivity , uncompensated materials ) and very high impurity concentration ( low resistivity ) , respectively . in both cases the equations
could be decoupled , and analytical solutions are possible .
so , if the formation of vacancy - impurity complexes is not so important , the main process responsible for the decay of vacancy concentration is divacancy production . in this case , the time evolution of the vacancy concentration is described by the equation : @xmath53 where : @xmath54 while the increase of the divacancy concentration is given by : @xmath55 the stationary solution for @xmath56 is given by : @xmath57 for n - type high doped si , the process described by eq .
( 4 ) is the most probable .
if @xmath58 is the initial concentration of impurities , and the initial concentration of complexes is zero , than : @xmath59 with : @xmath60 and with the stationary solution : @xmath61\ ] ] and @xmath62
in figure 1 , the absolute value of the cpd per unit fluence ( cpd ) induced by pions in silicon is presented as a function of the kinetic energy of the particle .
this curve has been obtained in the hypothesis listed in references @xcite . in the kinetic energy range between 100 to 300 mev ,
pion - nucleus interaction is resonant in all waves , the @xmath63 resonance is produced and a pronounced maxima in the cpd is observed at approximately 150 mev followed by a slight monotonic decrease at higher energies .
some local maxima are also presents at high energies , but with less importance .
the process of partitioning the energy of the recoil nuclei ( produced due the interaction of the incident particle with the nucleus , placed in its lattice site ) in new interaction processes , between electrons ( ionisation ) and atomic motion ( displacements ) has been considered in the frame of the lindhard theory .
the cpd multiplied by the fluence is the initial value of the concentration of vacancies and interstitials , and in the forthcoming discussion it is expressed , as specified , as atomic fraction . without loss of generality , will shall consider that the primary defects are produced by 150 mev kinetic energy pions .
the following values of the parameters have been used : @xmath27 = 0.4 ev , @xmath40 = 0.8 ev , @xmath44= 0.2 ev , @xmath45 = 0.2 ev , @xmath46 = 0.4 ev , @xmath28 = 10@xmath64 hz , @xmath65 = 10@xmath66 @xmath67 , @xmath30 = 10@xmath68 @xmath69 , the value of the parameter@xmath47 is taken 100 .
the time evolution of the concentration of vacancies , normalised to the concentration of vacancies created by irradiation is represented in figure 2 .
fluences in the range @xmath70 pions/@xmath67 have been considered .
the weight of the annihilation process in respect to interstitial migration to sinks increases abruptly with the fluence . at low fluences ,
up to @xmath71 pions/@xmath67 , the annihilation has a low importance . with a good approximation
, it could be considered that after 0.2 sec .
after irradiation the vacancy concentration is saturated . as could be seen from the figure , observable effects in @xmath72 appear for fluences higher than @xmath73 pions/@xmath67 .
experimentally , for silicon particle detectors irradiated with hadrons , some modifications in the electrical characteristics have been observed in the fluence range @xmath74 @xmath69 @xcite .
all the curves correspond to room temperature ( 20 @xmath75c ) annealing .
the interval of time after which the concentration of interstitials decreases @xmath47 times , given by eq .
( 17 ) with @xmath76 is a characteristic time of the processes in the first stage , and depends on temperature and fluence . in figure 3 , this characteristic time is represented as a function of fluence , in the fluence range @xmath70 pions/@xmath67 , for three temperatures of interest for high energy physics applications : -20 , 0 and + 20 @xmath75c .
it may be seen that , up to @xmath77 pions/@xmath67 , the characteristic time is independent on the fluence , and for higher fluences a decrease of the characteristic time is to be noted and a fluence dependence is obtained .
the initial concentration of vacancies for the second stage is @xmath48 previously calculated .
the second stage starts at a time equal to the characteristic time .
two limiting cases have been studied , low doping concentration ( high resistivity uncompensated material ) , and low resistivity , phosphorus doped silicon .
the formation and decomposition of divacancies is a good approximation for the first , while the second is related to @xmath78 complexes .
the time evolution of the concentration of vacancies when divacancy formation is considered is illustrated in figure 4a . as underlined in section 3 ,
the time scale of this process is much longer in respect to the first stage .
the origin of time in figure 4a is the beginning of the second stage , this means a characteristic time after the start of the annealing .
the curves in this figure are normalised to the concentration of vacancies remained after the first stage .
values of the initial concentrations of vacancies that correspond to fluences in the same range as in the previous analysis have been considered .
it could be observed that the time after which @xmath79 saturates is in the range @xmath80 sec . , and increases with the irradiation fluence .
all curves have been calculated also for 20 @xmath75c temperature .
the time evolution of divacancy concentration is represented in figure 4b , also in a similar manner with the vacancy concentration in figure 4a .
the case of high initial impurity ( phosphorus ) concentration , where divacancy formation is neglected in respect to complex formation , is presented in figures 5a and 5b respectively .
three initial concentrations for phosphorus in silicon ( expressed as atomic fractions ) have been considered : @xmath81 , @xmath82 , and @xmath83 , corresponding to 0.1 @xmath0 , 10 @xmath0 , and 1 @xmath1 resistivity respectively .
the time evolution of the concentration of vacancies is represented in figure 5a , while the corresponding one for the concentration of complexes in figure 5b , both normalised to the concentration of vacancies remained after the first stage .
an irradiation with @xmath77 pions/@xmath67 has been considered .
the origin of time in figures 5a and 5b is displaced in respect with the origin of the annealing , in the same way as in figs . 4a and 4b .
the curves corresponding to @xmath70 @xmath69 are superimposed on the same curve , with the exception of the tail at times longer than 10@xmath84 sec . where the values corresponding to higher fluences are slightly higher .
the time interval after which @xmath79 saturates is shorter than in the case of divacancy , and at room temperature is in the range @xmath85 sec . for the doping concentrations considered .
it shortens with the increase of impurity concentration .
some explicit considerations must be done about the formation of vacancy - impurity complexes .
the mechanisms supposed above can be used both for boron and for phosphorus impurities . in this case , the corresponding processes are @xcite : @xmath86\ ] ] and respectively : @xmath87\ ] ] while the complex formed by boron is unstable and self anneals bellow room temperature , the interaction between a @xmath2 and a @xmath88 leads to the formation of an @xmath89 centre which is stable in the same conditions @xcite .
interactions between interstitial oxygen ( another very studied impurity in the last time ) and free vacancies is described as a higher order process ( third order in the @xcite and fourth power in @xcite ) . if the process is stopped as a first order one , the time evolution of the concentrations is not different from the case of phosphorus , studied before . if two or more impurities that trap vacancies are considered as existing simultaneously in silicon , the system of coupled equations must be solved . only numerical solutions for particular cases
are possible .
if two or more impurities that trap vacancies are considered as existing simultaneously in silicon , the system of coupled equations must be solved . only numerical solutions for particular cases
are possible .
so , in the frame of this model , three characteristic times have been found for 20 @xmath75c self annealing in silicon : one in the order of 0.2 sec .
, corresponding to vacancy - interstitial annihilation and interstitial migration to sinks , one around 10@xmath84 sec . ,
related to vacancy - phosphorous formation and decomposition , and another one around 10@xmath90 sec .
, divacancy related .
the present model could be applied to the annealing of primary radiation defects induced in si by any other hadron , the difference coming from the value of the cpd .
the time evolution of the primary concentration of defects induced by pions after irradiation process has been modelled . in this model , vacancy - interstitial annihilation , interstitial migration to sinks , vacancy and interstitial impurity complex formation as well as divacancy formation have been considered . in the considered hypothesis ,
the time evolution of impurity concentrations has been decoupled into two steps , the first one involving vacancy - interstitial annihilation and interstitial migration to sinks , the second vacancy - complex and divacancy formation .
the equations corresponding to the first step have been solved analytically for a wide range of particle fluences and for a wide energy range of incident particles , and for three different temperatures : -20 , 0 and + 20 @xmath75c .
the approximations that permit to decouple the differential equations representing the time evolution processes in the second step have been studied , and the processes have been treated separately . the simultaneous consideration of more processes in this last step of the model is possible only numerically .
the predictions of the model are in general agreement with the experimental results , and suggest a possible connection with the self annealing mechanisms .
the authors are very grateful to professor gh .
ciobanu from the bucharest university for helpful discussions during the course of this work .
figure 4 : a ) time evolution of the concentration of vacancies , normalised to the concentration of vacancies remained after the first stage when divacancy formation and decomposition are considered , for different particle fluences , expressed in @xmath69 .
b ) same as ( a ) for the concentration of divacancies .
figure 5 : a ) time evolution of the concentration of vacancies , normalised to the concentration of vacancies remained after the first stage when @xmath78 complex formation and decomposition are considered .
the three curves correspond to different initial impurity concentrations , expressed in atomic fractions .
b ) same as ( a ) for the concentration of complexes . | the concentration of primary radiation induced defects has been previously estimated considering both the explicit mechanisms of the primary interaction between the incoming particle and the nuclei of the semiconductor lattice , and the recoil energy partition between ionisation and displacements , in the frame of the lindhard theory .
the primary displacement defects are vacancies and interstitials , that are essentially unstable in silicon .
they interact via migration , recombination , annihilation or produce other defects . in the present work ,
the time evolution of the concentration of defects induced by pions in medium and high resistivity silicon for detectors is modelled , after irradiation . in some approximations ,
the differential equations representing the time evolution processes could be decoupled .
the theoretical equations so obtained are solved analytically in some particular cases , with one free parameter , for a wide range of particle fluences and/or for a wide energy range of the incident particles , for different temperatures ; the corresponding stationary solutions are also presented . radiation damage , pions , atom displacements , kinetics of defects , annealing processes * pacs * : + 61.80.az : theory and models of radiation effects 61.70.at : defects formation and annealing processes and |
Subsets and Splits